Pogbagram

Just now, Anthropic announced a "ban" on Openclaw.

Many OpenClaw users have quietly received an email in their inbox. The email is short but contains a lot of information: Starting April 4th, Pacific Time, you will no longer be able to use your Claude subscription credit to access third-party tools, including OpenClaw.

Want to continue using it? There are two options.

You can either purchase additional usage bundles (offered at a discount by the official website) or provide your own Claude API Key and pay only for usage. There is compensation: a one-time monthly fee in credits, which will be converted into your monthly subscription fee. For a full refund, please find the link in the email.

Upon hearing the news, OpenClaw founder Peter Steinberger posted a poignant message on social media: "Board member Dave Morin and I tried our best to persuade Anthropic, and the best we could do was postpone it by a week."

"I made it with Claude."

The fact that OpenClaw became popular is itself an absurd story.

Peter Steinberger, the founder, is a legendary figure in iOS development and the creator of PSPDFKit. His coding skills should be unfathomable. However, when he created OpenClaw, he saw himself as a product manager.

The entire OpenClaw codebase, including the backend, frontend, CI/CD, tests, and documentation, is automatically generated by Claude Code. Peter himself is only responsible for describing the requirements in natural language.

I haven't written a single line of code myself.

The underlying technology of this tool is entirely based on Claude: extremely long context, agent tool calls, and multi-step inference planning. From its core to its framework, it is derived from Claude.

However, Anthropic still took action.

In fact, the feud between the two companies had already begun before this ban.

What was OpenClaw originally called? Clawdbot. Anthropic saw it and wasn't happy: "Your name sounds too much like Claude's. Change it." So it was first changed to Moltbot, and finally named OpenClaw.

After the name change, the two sides enjoyed a brief period of peace. Then OpenClaw became an internet sensation.

Managing inboxes, calendars, check-in assistance… users say this thing really works. One account can run ten agents simultaneously, 24/7 without interruption. A $200 Claude Max subscription has been calculated to offer up to $5000 worth of usable computing power.

This essentially means Anthropic is heavily subsidizing every heavy user. And OpenClaw users happen to be among the heaviest users. Putting them on Anthropic's servers is like creating a bloodsucker. So this conflict finally erupted today.

The founder fled to OpenAI

What makes the story even more intriguing is Peter's whereabouts.

At the beginning of the year, OpenAI poached him. What does this mean? OpenClaw, a tool heavily reliant on Claude, now has its founder working for Anthropic's most direct competitor. Anthropic must be experiencing a mix of emotions seeing this situation unfold.

In response to this controversy, Boris Cherny, the head of Anthropic Claude Code, posted four tweets in a row to personally explain the situation.

The first policy is announced directly:

"Starting at 12 PM PT tomorrow, Claude subscriptions will no longer cover the use of third-party tools such as OpenClaw."

Then, following the second point, the reason is explained:

"We have been working hard to meet the growing demand for Claude, and our subscription service was not designed for the usage patterns of third-party tools. Computing power is a resource that we need to manage carefully, and we prioritize our customers who use our own products and APIs."

In layman's terms: You're exploiting us too much; our machines can't handle it. Take care of your own people first.

Article Three addresses compensation:

"Subscribers will receive a one-time credit equal to their monthly fee. If you need more, you can purchase a discounted subscription package now. The link to apply for a full refund will be in your email tomorrow."

The fourth point concludes the discussion, and the art style becomes quite official:

"We want to consciously manage growth in order to serve our customers well in the long term. This change is a step towards that goal."

However, netizens are arguing fiercely. Blogger Yuchen Jin believes: "The Claude plan, priced at $200/month, can actually utilize up to $5,000 worth of computing power, which is an astonishing subsidy. Considering Claude's recent stability issues, and given Anthropic's current GPU computing power constraints, this adjustment may indeed be the right choice."

Blogger Peter Yang stated that Anthropic and OpenAI are now subsidizing heavy users who run multiple agents around the clock with a subscription price of $100-$200 per month, which is exactly the same as the crazy money burning by Uber and Lyft to grab market share back then.

As we all know, after both companies went public, ticket prices nearly doubled within a few years, and it took Uber a full 14 years from its founding to its first profit.

His assessment is that OpenAI and Anthropic are not far from going public, and once their financial data is released, these loss-making subscription plans will inevitably become unsustainable, either leading to price increases or data throttling. He also added insult to injury: "Running local models on the Mac Mini and Mac Studio is looking increasingly appealing now." (doge)

Some are thinking further ahead: "Codex is currently quite generous to third-party tools, given OpenAI's larger GPU reserves. The strategic divergence between the two companies is truly beginning to emerge. We'll see who can hold out the longest."

After wading through the road, the bridge was demolished.

Beyond cost, Anthropic has a deeper purpose in blocking OpenClaw.

Yes, Anthropic is simultaneously promoting Claude Code and Claude Cowork.

OpenClaw's control plane is built on chat applications such as WhatsApp, Discord, and Telegram. Users type in the chat box, and OpenClaw executes the commands. It runs 24/7 on your computer, automatically replying to messages, managing schedules, reading and writing files, and executing code.

But returning to the technical details, OpenClaw itself is not a model; it's a framework, a "shell." Its intelligence largely stems from Claude's modeling capabilities and Claude Code's CLI interface.

The most active users of OpenClaw mostly use Anthropic's API—as long as they can afford the token fee.

Anthropic has clearly recognized this. From its early protest against the name "Clawdbot" citing trademark and copyright issues, to the rapid release of four new features in the past two months—each one precisely targeting a core capability of OpenClaw:

• OpenClaw acts as a text proxy via WhatsApp messages, running on the desktop. Anthropic's response is Dispatch, a persistent thread from phone to desktop.
• OpenClaw uses Discord and Telegram as its control planes. Anthropic released Claude Code Channels, which bridges the two ends using the MCP protocol.
• OpenClaw provides full operating system access permissions, browser control, and application operation. Anthropic's corresponding versions are Computer Use and Claude Code, which share computer access within the same workspace.

But here's the interesting part: OpenClaw's success proves that users have a real and strong demand for this type of agent tool. This is a demand that Anthropic's own products couldn't meet at the time.

In other words, OpenClaw paved the way for Anthropic. Now that the path is clear, Anthropic is conveniently dismantling the bridge.

Under Peter Steinberger's tweet, a netizen commented: "Aside from all the previous experiences, I understand why you don't like them very much."

No one would argue with that statement.

Attached is a translation of the original email:

Hello,
From 12:00 PM Pacific Time / 8:00 PM BST on April 4th, you will no longer be able to use your Claude subscription credits to support third-party harnesses (including OpenClaw). You can still use these tools through your own Claude account, but at an additional cost, i.e., pay-as-you-go, which will be charged separately from your subscription.

Your subscription still covers all Claude products, including Claude Code and Claude Cowork. To continue using Claude to sign in to third-party harbors, please enable additional usage for your account. This policy will first be implemented with OpenClaw on April 4th, but it also applies to all third-party harbors and will soon be rolled out to more tools (learn more).

To facilitate a smoother transition, we will be offering a one-time credit for additional usage, equal to your monthly subscription price. Please claim this credit before April 17th. We will also be offering discounts on pre-ordered additional usage packages, up to 30% off.

We've been working hard to manage usage demands holistically, but these tools have placed significantly more strain on our systems. System capacity is a resource we need to manage carefully, so we must prioritize customers using our core products. You will receive another email from us tomorrow, at which time you can choose to get a refund for your subscription if you wish.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

Do you still use wired headphones?

On the same day that Apple's AirPods Max 2 was officially released, "wired headphones" stole the spotlight online, with hashtags like #WhyWiredHeadphonesAreRe-popular#, #WiredHeadphoneOutfitsAreTrending#, and #ReasonsForWiredHeadphonesBecomingPopular# trending on Weibo.

We then realized that although TWS true wireless earbuds have become the mainstream for the vast majority of people, wired earbuds still have a considerable number of fans.

I used to think wearing wired headphones was a sign of poverty, but now I think wearing wired headphones shows good taste.

Wired headphones are a trendy item.

The resurgence of wired headphones is not a sudden, overnight trend; in fact, it has been a revival that lasted almost a year.

A report from research firm Circana shows that after five consecutive years of declining sales, wired headphones suddenly rebounded in 2025, with annual sales growth of 3%, and accelerated in the second half of last year, with sales surging by 10%. In the first six weeks of 2026, sales even soared to 20%.

Meanwhile, the entire headphone market is experiencing a significant contraction. According to the latest report from Lotoo Technology, in 2025, the sales volume of headphones and headsets in China will decrease by 6.9% year-on-year, marking the first decline since 2019.

The release of this data, coupled with attention and reports from some media outlets, has resonated with many people, making it the hottest digital topic this week.

We originally thought that when Apple removed the 3.5mm headphone jack from the iPhone 7 and launched the true wireless earphone product AirPods, wired headphones were doomed.

Unexpectedly, ten years later, wireless headphones are no longer selling well, while wired headphones are bucking the trend and becoming increasingly popular.

Unlike smartphones and watches, headphones inherently possess "OOTD" (Out-of-Site-Time) attributes, and recent TWS (True Wireless Stereo) headphones have clearly been moving in this direction. Besides sound quality and functionality, becoming a "trendy item" is also a way to break through the competition.

As we entered the 2020s, the world was enveloped in a retro sentiment, with various "yesterday's products" becoming the latest fashion items: CCD cameras, phone straps, iPhone 4, and now wired headphones.

This wave of cultural renaissance was primarily driven by fashion trends and the influence of celebrities.

In the past two years, South Korean female idols' street photos of commuting to and from work and at airports have been dominated by AirPods Max. In the past two years, they have generally been wearing wired headphones.

Jacob Elordi, the actor who rose to fame with "Wuthering Heights" and "Euphoria," is particularly fond of wired headphones, leaving behind many memorable street photos—sometimes he even wears these headphones to pose for a photo, without any device connected to the other end.

At New York Fashion Week 2025, American singer and actress Dove Cameron even wore wired headphones as part of her hair accessory, sparking heated discussions in the fashion world.

Interestingly, the headphones these celebrities used for their "photo shoots" were almost all very simple white wired headphones, most likely the classic Apple EarPods.

This brings us to Apple's fashion sense, which was 20 years ahead of its time. Before EarPods, headphone cables were mostly black or gray. Apple made white headphone cables, originally just to match the white iPod, but they quickly realized that the unique color of the headphones would make it immediately obvious that you had an iPod in your pocket.

This led to the creation of the classic iPod silhouette advertisement, which itself became an important visual symbol in popular culture, giving the white headphone cable even more aesthetic significance.

TWS earbuds in the shape of earbuds are too small and need to be stored in the charging case, lacking space for "creative" designs. Wired earbuds with a long cable, on the other hand, naturally start to be met with disapproval.

Generation Z, who love surfing the internet, may not have experienced the "vintage" era of CCD cameras, but most of them used wired headphones when they were young. This resurgence can quickly activate their usage experience and memories, and the resonance is actually deeper.

Trends may be fleeting, but I don't think wired headphones will quickly become obsolete, for a simple reason— they are genuinely useful.

Wired = "Reliable"

Wireless headphones have only drawbacks except for being wireless, while wired headphones have only advantages except for being wired.

I personally disagree with this statement. Wireless earbuds are ready to use right out of the box and easy to store. Compared to wired earbuds, which can get tangled and require checking the headphone jack, the experience is much smoother. Features like active noise cancellation and spatial audio are also generally only available in wireless earbuds.

However, in some situations, I still prefer the simplicity and directness of wired headphones.

First of all, whether it's a Mac for work or a Windows for entertainment, their support for Bluetooth headsets is almost always very unreliable.

Even when using AirPods, I always encounter inexplicable disconnections when connecting to my Mac. Even though the headphones are connected to my computer via Bluetooth, I have to reconnect after enabling AirPlay. Furthermore, my AirPods Max always activate the microphone when connected to my Mac, resulting in a complete loss of sound quality.

In addition, I wear headphones to listen to music for a relatively long time at work. Wireless headphones tend to run out of battery easily, and the wearing experience is not as comfortable as semi-in-ear headphones. I also don't need to use active noise cancellation in the office.

Microphones remain a difficult area for wireless headphones to conquer. When playing games using Windows Voice Chat, I usually use wired headphones because the microphone is generally located below the lips, providing clear sound pickup and requiring no additional setup.

I'm also an ASMR enthusiast and like to listen to some relaxing videos before bed. If I'm wearing AirPods and don't put them away properly before falling asleep, it will not only put a strain on my ears, but I might also find that both earbuds are gone when I wake up.

Most importantly, I'm using a pair of Onda headphones that cost around 10 yuan. The sound quality is excellent, and even if it breaks or gets lost, I can easily buy another pair.

These scenarios, though based on my personal experience, are actually shared by friends, netizens, and even professional musicians.

In terms of "reliability," wired headphones are still too authoritative.

The wireless era is inevitable, but wired headphones will always exist.

The removal of the 3.5mm headphone jack from smartphones, ushering in a full-fledged "wireless era," is not as negative as some people think. In fact, I think it has some positive significance because it has indeed popularized the easy experience of wireless headphones.

Ten years have passed, but the so-called "wireless era" is still much the same as when AirPods first came out: Apple's top-of-the-line AirPods Max still requires a wired connection to listen to Apple's own lossless audio; connecting wireless headphones to a computer still presents a host of problems; and buying a decent pair of wireless headphones is not cheap, plus you have to be careful to buy products from the same ecosystem as your phone…

The key point is that, besides mobile phones, many other products can still use 3.5mm wired headphones: Apple's cheapest laptop, the MacBook Neo, can, as can the Switch 2 and various game controllers, and camera monitoring devices.

What many people may not know is that when you buy a pair of headphones now, they usually come with an audio cable, allowing you to "turn" them into wired headphones with better sound quality. This seems to confirm that we cannot completely leave "wired" headphones behind.

It seems that the reason why everyone pursued wireless headphones a few years ago was simply because their most commonly used mobile phones lacked a dedicated interface for audio services.

The reason why true wireless earbuds sold so well in the past few years was more like a "from zero to one" transformation – more and more users who originally only had wired earbuds started to buy wireless earbuds, and the whole market flourished.

Nowadays, almost everyone owns a pair of TWS earbuds. The only difference between new ones is that they have better noise cancellation or sound quality, so there's no need to buy new ones. On the contrary, once people have experienced the wireless feature, they start to appreciate the advantages of wired earbuds.

For most people, perhaps "dual-wielding" is the ideal state.

I still have high hopes for wireless headphones and the entire "wireless era." After all, wireless headphones are more powerful, allowing them to go beyond just listening to music. In the future, they may very well become an AI gateway, and the freedom from cables has always been a pursuit of human technological progress.

However, I also believe that as long as electronic products haven't completely eliminated all interfaces, wired headphones won't truly disappear—and even after that, wired headphones will continue to exist, only becoming a symbol worn on the body to express oneself.

Cover image source: Teen Vogue

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

The Nürburgring lap time rankings are among the few lists that directly compare cars with different powertrains and engineering approaches using the same set of standards.

Previously, the Porsche 919 Hybrid Evo was in first place with a time of 5 minutes 19.546 seconds, a hybrid; the Volkswagen ID.R was in second place with a time of 6 minutes 05.336 seconds, a pure electric vehicle; the Xiaomi SU7 Ultra prototype was in third place, a pure electric vehicle; and the Lotus Evija X was in fourth place, also a pure electric vehicle.

But today, this list has been rewritten by a pure gasoline-powered performance car from the United States.

The Ford GT Mk IV surpassed Xiaomi with a time of 6 minutes 15.977 seconds, rising to third place and becoming the fastest all-fuel car on the Nürburgring, as well as the fastest production car on the Nürburgring.

In a list dominated by hybrid and pure electric vehicles, the Ford GT Mk IV, as a pure gasoline car that insists on the pure roar of the internal combustion engine, has carved out a path for traditional gasoline cars and secured its own place with its extremely hardcore mechanical quality and top-notch engineering tuning.

Lei Jun also retweeted the post to congratulate Ford.

The GT Mk IV is the final evolution of the third-generation Ford GT, positioned as a pure track car, with a limited production run of 67 units worldwide.

Its power comes from a twin-turbocharged EcoBoost V6 engine that was extensively developed for this car. Engineers overcame many limitations of mass-produced engines, increasing the maximum power to over 800 horsepower.

This engine is paired with a custom-designed racing-grade transmission. Its calibration logic is almost entirely focused on track efficiency, disregarding the shift smoothness emphasized in civilian vehicles.

Each gear engagement carries the most primal mechanical impact, and its sole purpose is to convert engine output into a faster and more accurate track response as completely as possible.

However, brute force alone is far from enough on the Nürburgring. The Nürburgring is over 20 kilometers long, with an elevation change of over 300 meters, and is dotted with more than a hundred extremely challenging curves. The road load continuously accumulates and changes drastically during high-speed driving. Traditional shock absorbers with fixed parameters set before departure are simply unable to cope with such complex and diverse conditions.

To this end, Ford equipped the GT Mk IV with an adaptive spool valve suspension system provided by Multimatic. This system allows each shock absorber to sense road undulations and vehicle posture in real time and complete independent, continuous damping adjustments in milliseconds.

Ford has also pushed the limits in terms of aerodynamics. The extended wheelbase directly improves the vehicle's stability at ultra-high speeds, while the exaggerated carbon fiber tail styling completely relegates "aesthetics" to a secondary position. Almost every detail on the body serves the same goal: to maximize downforce while maintaining a balance between drag efficiency and overall performance.

Every detail on the car body is the optimal solution after countless wind tunnel tests. While extracting downforce, they also maintain a delicate balance of wind resistance efficiency.

Of course, even the most powerful machines need to be pushed to their limits by humans. The driver who set the record in this Ford GT Mk IV was top-tier driver Frédéric Vervisch, with extensive experience in the Nürburgring 24 Hours endurance race. His perfect coordination with this mechanical beast ultimately secured a lap time of 6 minutes and 15.977 seconds, a number that remains on the lap time leaderboard.

The GT Mk IV's performance was not a sudden, accidental breakthrough.

Back in December 2024, the Ford Mustang GTD, equipped with a 5.2-liter supercharged V8 engine, had already achieved a time of 6 minutes and 57.685 seconds, officially breaking the 7-minute barrier and becoming one of the few mass-produced sports cars in the world to do so.

Going back further, Ford's investment in extreme racing can be traced back to 1966, when the Ford GT40, equipped with an FE series V8 engine, won the race, ending Ferrari's almost unshakeable dominance at Le Mans.

Ford went on to win Le Mans for four consecutive years. To this day, it remains the only American automaker to have won the highest class of the Le Mans series.

On the Formula 1 circuit, Ford, as an engine manufacturer, has helped participating teams win 174 Grand Prix victories and 10 Constructors' Championships, making it the third-most successful engine supplier in F1 history.

This path of accumulating technology on the brutal racetrack and then transferring the results to civilian applications has long been ingrained in Ford's brand DNA. The EcoBoost series engines, which have won numerous international awards and are widely used in Ford's mass-produced models, are the most direct manifestation of this "racetrack technology feeding back into civilian applications."

However, while Ford represents the pinnacle of gasoline-powered vehicle technology on the racetrack, Ford in the civilian consumer market, especially amid the global wave of electrification in the automotive industry, is currently undergoing a rather difficult period of adjustment.

In the first quarter of this year, Ford's overall sales fell by 8.8% year-on-year, a decrease of approximately 44,000 vehicles.

The pressure on the electric vehicle business is particularly evident. Affected by factors such as the cancellation of the US federal tax credit, increased consumer hesitancy, and the discontinuation of some models, sales of Ford's previously highly anticipated electrified flagship models have seen a significant decline.

Sales of the Mustang Mach-E plummeted by more than 60%; the once highly anticipated all-electric pickup truck F-150 Lightning saw a sales decline of 71.3%; and the commercial electric vehicle E-Transit has almost come to a standstill.

Currently, the real pillars supporting Ford's profits are still its more traditional products. Financial reports show that large-size gasoline SUVs, such as the Expedition and Explorer, have achieved sales growth of approximately 30% against the trend; while sales of the traditional gasoline performance car, the Mustang, have increased by more than 50%.

On one hand, Ford is struggling to advance its electrification transformation at the sales level; on the other hand, it is pushing pure gasoline technology to its limits on the racetrack. These two issues occurring simultaneously give the impression that Ford is trapped at a crossroads, a crossroads of past glories and future uncertainty.

However, Ford has officially announced to the world that this limited edition of 67 GT Mk IVs is the final chapter of the Ford GT series.

After pushing the limits of pure gasoline-powered vehicles to an unprecedented and perhaps unparalleled level, Ford is now moving toward new technological directions.

In 2026, Ford officially returned to the F1 (Formula One Championship) circuit, and they began a deep collaboration with the Red Bull team, which currently dominates the paddock, to jointly develop a new generation of hybrid systems.

Although still on the track, the traditional roar of pure internal combustion engines has become history, replaced by highly complex hybrid systems and energy recovery technologies.

Since the times have changed, Ford has had to accept this shift and switch its technological foundation to a new track.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

News worth watching even on the weekend

Apple reportedly made a massive DRAM purchase to prevent competitors from acquiring it.

According to a report by the source "Jukan", Apple has almost "sweeped up all available mobile DRAM on the market" at an extremely high premium, and is even willing to pay the price of a loss in operating profit margin for this.

The source emphasized that Apple's move was not simply to meet its own regular inventory needs, but rather aimed at artificially inflating DRAM market prices to prevent other smartphone manufacturers from obtaining enough memory chips.

Analyst Ming-Chi Kuo previously released a research report pointing out that Apple is leveraging its supply chain power to lock in production capacity during the memory chip price increase cycle, and plans to absorb cost pressures in exchange for a larger market share.

Although rising memory costs will inevitably squeeze iPhone's gross margin, Ming-Chi Kuo points out that Apple's core strategy is to lock in chips, absorb costs, and seize market share during periods of market turmoil, and then make up for the losses through growth in its service business.

In addition, to facilitate marketing, Apple currently plans to keep the starting price of the iPhone 18 series, which will be released in the second half of 2026, as unchanged as possible.

Xiaomi officially announced a price increase for its mobile phones; Lu Weibing stated that the increase in memory prices far exceeded expectations.

Xiaomi officially announced yesterday that, due to the continued sharp rise in the prices of key components such as global memory chips, it will adjust the suggested retail price of some of its products starting April 11.

This price adjustment involves three REDMI models: the suggested retail price of the REDMI K90 Pro Max will increase by 200 yuan; the REDMI Turbo 5 will no longer have its Spring Festival special offer, but the 512 GB version will continue to receive a 200 yuan subsidy; the REDMI Turbo 5 Max will also no longer have its Spring Festival special offer, but the 512 GB version will also retain a 200 yuan subsidy.

In response, Lu Weibing, partner and president of Xiaomi Group, posted on Weibo, saying, "The price increase of memory in this round has indeed far exceeded expectations. The price of the same version of memory has soared by nearly four times compared to Q1 of last year… We have no choice but to make a slight increase in the retail price of some models or restore the original price. We hope everyone understands."

Wei Siqi, General Manager of Xiaomi China Marketing Department, also stated that the company had been making every effort to suppress the impact of memory price increases on terminal prices, but the speed and magnitude of this round of price increases exceeded expectations. In order to ensure normal product supply and quality stability, the company ultimately decided to adjust the prices of some products.

According to Caijing Magazine, Xiaomi internally expects its mobile phone sales to decline by 13% this year and is working to adjust its offline store strategy, shifting its revenue focus to large home appliances while halting offline store expansion.

Power banks will face the strictest national standard in history, to be mandatory next year.

According to CCTV News, the mandatory national standard GB 47372-2026 "Technical Specification for Safety of Mobile Power Banks", which was organized and formulated by the Ministry of Industry and Information Technology, was officially released yesterday (April 3) and will be officially implemented on April 1 next year.

The new national standard covers two product categories: "power banks" and "outdoor power supplies".

In terms of safety, the standard introduces a needle penetration test to clearly improve the safety protection capabilities of power banks in extreme scenarios such as high temperature, overcharging, and compression. At the same time, it adds a cycle aging test to reduce the risk of internal short circuits after long-term use, and requires products to indicate the recommended safe service life to remind consumers to replace aging products in time.

In addition, the new national standard promotes product unique coding management, allowing consumers to query core information such as battery brand through the code; in terms of production and manufacturing processes, it clearly puts forward requirements for the control of raw materials and production processes, improving the level of safety from the source.

Experts say that after the implementation of the new national standard, power banks that have been purchased and passed the old 3C certification can still be carried on board as long as they comply with the current relevant regulations of the Civil Aviation Administration of China.

The drafting team for this new national standard is large, encompassing more than 30 companies and academic institutions, including the China Electronics Technology Standardization Institute, CATL, Huawei Terminal, Sunwoda, Anker Innovations, Xiaomi, OPPO Guangdong Mobile Communications, BYD Lithium Battery, Lenovo, and Bull Group.

Ford breaks Nürburgring North American vehicle record, Xiaomi's ranking slips, Lei Jun posts congratulatory message.

On April 3, Ford China announced that the Ford GT Mk IV set a new lap record at the Nürburgring Nordschleife with a time of 6:15.977, becoming the fastest lap time ever achieved by an American manufacturer at the Nordschleife.

In response, Xiaomi CEO Lei Jun immediately reposted the message on Weibo and commented "Congratulations to Ford," along with a thumbs-up emoji. Lei Jun added in the comments that the Xiaomi SU7 Ultra is currently ranked 4th globally on the official Nürburgring lap time list, "which is nothing to be ashamed of."

Previously, the Xiaomi SU7 Ultra garnered widespread attention for its Nürburgring lap time. Now, with the Ford GT Mk IV breaking the record, Xiaomi's ranking on that list has dropped one place.

Slack's decision to cut off services in Greater China has sparked strong backlash from users.

According to The Information, Slack recently suspended all Workspaces with billing addresses in mainland China, Hong Kong, and Macau, affecting both paid and free users.

After being deactivated, users will not be able to access historical messages, files, channels, and workflows. Slack will also initiate a 90-day data deletion countdown. Upon verification, iFanr's own Slack Workspace has also been deactivated.

Last November, The Information first reported that Slack had sent notices to users in affected regions, requiring them to migrate their accounts to Alibaba Cloud by February of this year, or their subscriptions would not be renewed.

However, the actual implementation has sparked strong dissatisfaction among many users. Several users in Hong Kong on Reddit reported that they never received any prior notification, or that the notification emails were categorized as spam by the system.

One user wrote, "We are completely locked out and have no access to any files, customer communication records, or historical data. Slack's handling of this feels like extortion—they cut off our access without giving us any window to export the data."

Slack Support responded with a standard template, confirming that "the Workspace has been affected by the previously notified service changes," but did not provide a specific data export process or timeline.

The notifications themselves also showed a clear difference in treatment: larger paying customers received migration guidance from Alibaba Cloud, while smaller teams received service termination notices directly without being offered migration options.

The Ministry of Industry and Information Technology (MIIT) has issued a warning: iOS versions 13 through 17.2.1 contain serious vulnerabilities and are advised to upgrade as soon as possible.

According to Jiemian News, the Ministry of Industry and Information Technology's Cybersecurity Threat and Vulnerability Information Sharing Platform (NVDB) recently detected that attackers are using exploit tools targeting Apple's terminal products to carry out cyberattacks, which could lead to serious harm such as information theft and system control. The affected devices include iPhones, iPads, and other Apple terminal products running iOS versions 13.0 to 17.2.1.

The Ministry of Industry and Information Technology (MIIT) recommends that relevant users complete risk assessment as soon as possible, fix vulnerabilities by upgrading system versions or installing patches, and refer to Apple's official security update announcements.

Vibe Coding app Anything was removed from the Apple App Store and then relisted.

The official account of the Vibe Coding app "Anything" has just confirmed that Apple has relisted the Vibe Coding tool Anything on the App Store. To celebrate its relisting, Anything has announced a $5,000 weekend hackathon with 20,000 credits as prizes.

Previously, Apple removed Anything from the App Store for violating Article 2.5.2 of the App Store Review Guidelines.

The terms stipulate that apps should be self-contained within their own suite and must not download, install, or execute code that introduces or alters the app's features or functionality. Anything allows users to describe their needs in natural language on their iPhones, with AI generating code in real time and running it on the device, directly conflicting with these terms.

In response to the removal, Anything posted a statement on the X platform: "Breaking news: Apple is panicked about Vibe Coding and has removed Anything from the App Store, so we've moved app development to iMessage. We wish Apple a successful removal."

It's worth noting that Apple itself is not against AI-assisted programming—Xcode 26.3, released in February of this year, already integrates Claude Agent with OpenAI Codex.

What Apple is really resisting is users generating and executing new code outside of the store's review process through third-party apps. This controversy is essentially a battle for platform control.

Related reading: Apple, Vibe Coding's strictest father

Alibaba's 1000 Questions 3.6 tops the list of domestic AI programming challenges.

On April 2, Alibaba released its new generation large language model, Qwen 3.6-Plus. The day after its release, the model ranked second globally in the React category of the Code Arena leaderboard under LMARaena, a world-renowned blind testing platform for large language models, with a score of 1452, becoming the highest-ranking domestic large language model on the list.

According to the rankings, Qwen 3.6-Plus ranked second with 1452 points, second only to Anthropic's Claude-Opus-4.6-Thinking (1540 points), leading GPT-5.0-High (1448 points) by 4 points, and surpassing Gemini 3.1 Pro Preview (1440 points) by 12 points.

In Code Arena's comprehensive ranking of AI programming capabilities, Qwen 3.6-Plus also ranked first among domestic models. Based on these achievements, Alibaba rose to fourth place in the global AI lab rankings, behind only Anthropic, OpenAI, and Google.

LMArena employs a real-user blind testing and real-time adversarial ranking mechanism, and is considered a highly credible performance evaluation platform in the AI ​​field.

Rockstar's parent company reportedly disbands its AI team, bringing seven years of research to an abrupt halt.

According to Insider Gaming, Take-Two Interactive, the parent company of GTA 6 developer Rockstar, made significant changes to its AI division yesterday, laying off an unspecified number of employees, including the head of AI.

Luke Dicken, former head of AI at Take-Two, confirmed the news on LinkedIn: "I regret to inform you that my team and I have lost our jobs at T2."

He said he would release a more detailed retrospective article in a week and called on industry professionals to help team members find new jobs, saying that these people "have spent seven years developing cutting-edge technology and are good at combining innovation with product design to create systems that empower the entire development process."

Take-Two has not yet publicly disclosed the specific scale and reasons for the layoffs, but it is widely interpreted as a signal that the gaming industry is scaling back its investment in AI research and development. It's worth noting that Take-Two CEO Strauss Zelnick previously stated publicly that AI "will never" be able to create a game like GTA 6, and has long endorsed the human development model.

This is not the first time Take-Two has laid off a large number of employees in recent years. In 2024, the company cut about 5% of its workforce and canceled several projects in development to improve financial efficiency. This restructuring of the AI ​​division occurred several months before the release of GTA 6, making the timing particularly sensitive.

Honor and JD.com reach a comprehensive strategic partnership, with a three-year target of exceeding 100 billion yuan.

Recently, Honor and JD.com signed a strategic cooperation agreement to establish a comprehensive strategic partnership. According to the agreement, the two parties will carry out in-depth cooperation in three core areas: product co-creation, user co-management, and ecosystem sharing, and set a goal of total cooperation exceeding 100 billion yuan over three years.

In terms of product co-creation, the two parties will cover the entire lifecycle of multiple product categories, including mobile phones, all-scenario products, AIoT, and robots. The cooperation includes the launch of strategic new products, joint IP and C2M customized projects, and scenario-based product co-creation targeting specific groups such as delivery riders, couriers, and gamers.

At the user co-management level, both parties will integrate full-link marketing resources, enhance user experience through joint membership operations and precise audience targeting, and conduct in-depth cooperation in e-sports, sports and health and other scenarios.

In terms of ecosystem sharing, the two parties will leverage Honor's edge-side big model capabilities and JD.com's JoyAI big model, JoyInside and other AI ecosystems to jointly create innovative experiences in scenarios such as product guidance, lifestyle services and financial management.

In the robotics field, JD.com's offline stores will pilot the deployment of Honor robots for customer acquisition and sales guidance. Furthermore, the two companies will leverage Honor's terminal products and JD.com's Joybuy platform to collaboratively expand into overseas markets and further deepen cooperation in areas such as omnichannel O2O, instant retail, and fintech insurance.

Shanghai Jiao Tong University and Ant Health jointly establish a medical AI laboratory.

The School of Artificial Intelligence at Shanghai Jiao Tong University and Ant Health recently signed a cooperation agreement in Shanghai. The two parties will jointly establish the "AI4HealthCare Joint Laboratory" to tackle joint research and development and clinical application of specialized intelligent agents in the medical field.

The joint laboratory will leverage Shanghai Jiao Tong University's leading advantages in basic research, algorithm models, and academic innovation, combined with Ant Health's accumulated scenario resources, technical capabilities, and industry resources in the healthcare field, to jointly explore innovative application models of AI in disease prevention, health management, and assisted diagnosis and treatment.

Both parties will also collaborate on artificial intelligence disciplines and clinical specialties in affiliated hospitals, focusing on advancing key technology research and development for intelligent medical agents and their clinical applications. The relevant findings will first be implemented on the Ant Financial App.

McKinsey senior partner: AI is "flattening" your company, and management levels will be drastically reduced.

According to Business Insider, McKinsey senior partner Alexis Krivkovich recently stated on The McKinsey Podcast that AI is giving business leaders "superhuman management capabilities," enabling them to manage larger teams and thus driving the flattening of corporate organizational structures.

Krivkovich points out that over the past decade, companies have added an average of at least one to three management levels between the CEO and frontline employees. This not only brings high human resource costs but also slows down decision-making efficiency—each level means an additional hurdle that requires "signature approval."

The intervention of AI agents is replacing these intermediate nodes, directly assuming the functions of decision support and information transmission.

Eno Reyes, co-founder and CTO of Factory, an AI-native software development platform, told Business Insider in March, "Your organizational chart will likely start to flatten out horizontally."

Factory has already deployed autonomous programmable agents for consulting firms such as KPMG and technology companies such as NVIDIA and Adobe.

IBM is also pursuing a similar strategy. Mohamed Ali, the company's senior vice president, revealed that IBM plans to deploy "digital employees" in conjunction with its 150,000 human consultants, and anticipates that managing AI will be fundamentally different from managing human employees in the future.

There will be a dedicated system to manage these things, and a dedicated system to set up guardrails.

Starting at 39 yuan/month, Xiaomi MiMo launches a subscription-based Token plan.

Yesterday, Xiaomi officially launched the Xiaomi MiMo Token Plan, offering subscription-based token packages to global developers and AI users. The packages are divided into four tiers:

• The Lite plan costs 39 yuan/month and provides 60 million credits, enough to perform approximately 120 medium to complex tasks.
• The Standard package costs 99 yuan per month and provides 200 million Credits, which can be used to execute approximately 400 tasks.
• The Pro plan costs 329 yuan per month and provides 700 million Credits, which can be used to execute approximately 1,400 tasks.
• The Max plan costs 659 yuan per month, provides 1.6 billion credits, and can execute approximately 3,200 tasks, targeting high-intensity usage scenarios around the clock.

Regarding tool compatibility, the MiMo Token Plan has been adapted to mainstream AI programming tools and development platforms such as Claude Code, Cline, Kilo Code, OpenCode, Hermes Agent, and CodeBuddy Code. Xiaomi stated that automatic renewal and more flexible usage management features are under development.

The three major food delivery platforms were summoned for talks and ordered to implement new food safety regulations by June 1st.

According to CCTV News, the State Administration for Market Regulation recently held an administrative guidance meeting for food delivery platform companies in Beijing, requiring Meituan, Taobao Flash Sale, and JD.com to conduct self-inspection and rectification as soon as possible and strictly implement the "Regulations on Supervision and Management of Online Catering Service Operators' Implementation of Food Safety Responsibilities," which will be implemented soon.

The meeting required all platforms to align themselves with the "Regulations" item by item and fully fulfill their platform responsibilities in five dimensions: qualification review, management and safeguards, public disclosure, emergency management, and collaborative supervision.

Meanwhile, taking June 1st, the date when the "Regulations" officially come into effect, as the deadline, we will work backwards from the deadline and comprehensively review the existing systems, processes, staffing, and technical support.

The regulators also emphasized that the three major platforms must strictly control the review, management, and delivery processes, strengthen their management capabilities through technological means, and proactively integrate into the government's regulatory system, shifting from "passive response" to "proactive engagement."

Specific measures include promoting the implementation of "Internet + transparent kitchen" and encouraging food delivery riders to participate in food safety supervision, so as to jointly build a social governance pattern for food safety.

Didi opens up its ride-hailing skills, supporting "Lobster" ride-hailing.

Didi Chuxing announced yesterday that it has officially opened its ride-hailing skill "didi-ride-skill" to "OpenClaw" users, providing them with full access to travel services.

didi-ride-skill packages and opens up Didi's underlying travel capabilities accumulated over many years, covering functions such as address parsing, price estimation, route planning, driver matching, order tracking, and nearby search.

After installation, you can directly initiate a ride request in the Lobster chat box using natural language, without having to switch to a separate app. Skill will handle all operations, including address confirmation, vehicle selection, order placement, and order status check.

didi-ride-skill also supports remembering user preferences, such as "take a ride-hailing service to work and a special offer ride after get off work" or binding "go home" to a fixed address for faster response in subsequent requests.

Didi also disclosed usage data for its AI-powered ride-hailing service: In the past week, the number of users using Didi's AI-powered ride-hailing service has increased 37 times compared to the beginning of the year, with users born after 2000 accounting for more than 40%, making them the main user group for the current AI-powered travel function.

GitHub: github.com/didi/didi-ride-skill

Say goodbye to gacha games! Alibaba releases Wan2.7-Video, allowing users to edit videos with just one sentence.

On April 3, Alibaba Tongyi Labs officially announced the release of the Wan2.7-Video video generation series models, which cover four professional sub-models: text-based video (Wan2.7-t2v), image-based video (Wan2.7-i2v), reference-based video (Wan2.7-r2v), and video editing (Wan2.7-videoedit).

• Video editing : Supports partial or overall modification of the image through natural language commands, covering operations such as adding or deleting elements, replacing objects, modifying attributes, and switching scenes, seasons and visual styles (such as realism to watercolor, live-action to clay animation);
• Plot and shooting adjustments : Without changing the original identity and scene, it supports modifying the character's lines (automatically matching lip movements and timbre), actions, and shooting parameters such as camera position, angle, shot size, and focal length;
• Creative replication : Supports reference migration of action sequences, camera movements, visual styles, particle effects, and other dimensions, with a maximum of 5 video subjects to be specified at the same time;
• Story continuation : Introducing a combined mechanism of video continuation and end frame control, which can maintain dynamic continuity while specifying the ending scene, solving the "braking" problem common in traditional first and last frame solutions;
• Character control : Supports multimodal reference locking of character appearance and voice, supports more than 40 subdivided expressions, and improves the realism of the sound and the quality of audio-visual synchronization. The related audio capabilities are technically supported by the Tongyi Labs voice team.
• Camera movement capabilities : Supports dozens of basic camera movements as well as composite camera movement schemes such as Hitchcock zoom, 360-degree surround, FPV drone dive, and one-shot-through.

Ant Financial releases professional AI agent DTClaw

Ant Financial officially announced yesterday that its professional AI agent product, DTClaw, has entered the internal testing phase. Positioned as a "professional assistant," this product aims to provide 24/7 online dedicated AI agent services to professionals such as financial experts, marketing experts, and data experts.

According to reports, DTClaw focuses on "being an expert from the start," with hundreds of built-in professional skills covering high-value scenarios such as financial management, investment, data analysis, R&D and testing. It also deeply integrates with enterprise office, finance, sales and marketing systems, supporting workflow automation.

According to the official statement, users can get started with one-click deployment, without having to configure from scratch, which can save 50% of token consumption.

In terms of safety, DTClaw has created the CARLI safety model, which establishes a safety barrier for the operation of intelligent agents by equipping AI with "brakes" and "black box" mechanisms.

Manner coffee prices rise

According to Jiemian News, Manner Coffee recently announced a price adjustment on its official ordering mini-program, stating that starting April 7th, the price of each cup of SOE coffee will increase by 5 yuan, and the price of the popular "Wizard of Green" coffee will also increase by 5 yuan.

Manner Coffee stated that the nationwide price increase is primarily due to the use of higher-quality coffee beans to provide consumers with a superior flavor experience. The company also emphasized that its long-standing "bring your own cup and get a NT$5 discount" promotion remains unchanged.

The AI-generated short drama "Peach Blossom Hairpin" was completely removed from all platforms due to AI face-swapping, despite having over 40 million views.

Yesterday, ByteDance's subsidiary, Hongguo Short Drama, announced the complete removal of the short drama "Peach Blossom Hairpin" from its platform and a 15-day suspension of all episodes uploaded by the production company.

The incident began on the evening of March 30, when Hanfu makeup blogger "Baicai Hanfu Makeup" publicly accused the producers of "Peach Blossom Hairpin" of using her personal photos without authorization, generating a character image through AI technology, and portraying the character as a negative figure who is "greedy and lustful".

According to Phoenix Network Technology, "The Peach Blossom Hairpin" is an AI-generated short drama in the ancient style, which has been viewed more than 40 million times on relevant platforms and collected by more than 28,000 people.

On April 2, the Actors Committee of the China Federation of Radio and Television Social Organizations issued a statement pointing out the current chaos of infringement such as AI face-swapping and voiceprint cloning, and demanding that online platforms strictly implement their main responsibility for content review and establish a long-term mechanism for verifying the authorization of AI performance content.

In its statement, Hongguo Short Drama said that after receiving the complaint, the platform immediately began to review and verify the relevant information and materials of the production company. After careful comparison and verification, the platform determined that the production company had violated the relevant content compliance regulations, constituting a breach of contract, and therefore made the decision to remove the content and suspend its uploading.

The statement also indicated that for producers who violate regulations or breach contracts, the platform will take measures such as removing content, terminating cooperation, and pursuing relevant legal responsibilities, depending on the severity of the violation.

LEGO partners with four football stars to launch the Editions series in preparation for the 2026 World Cup.

The LEGO Group announced yesterday a collaboration with Cristiano Ronaldo, Kylian Mbappé, Lionel Messi, and Vinicius Junior to launch a new LEGO® Editions set in celebration of the 2026 FIFA World Cup.

This release covers three product lines: "Classic Goal Moments," "Football Legends," and "Messi's Celebration Moments."

• Classic Goal Moments (Suitable for ages 10 and up): Ronaldo (43012), Mbappe (43013), Messi (43011), Vinicius (43027) – four designs, 490-510 pieces each, priced at 279 yuan each;
• Football Legends (Suitable for ages 10 and up): Ronaldo (43016) and Messi (43015) are two display-scale minifigures, each with two build options, priced at NT$699 each;
• Messi's Celebration Moment (43018, suitable for ages 14 and up): 1427 pieces, can be used as a 3D wall decoration, priced at 1599 yuan.

"Cold War 1994" returns to the big screen, scheduled for release on May Day.

The crime action film "Cold War 1994," directed by Longman Leung, has officially announced its release date for May Day, along with a trailer and poster. This marks the return of the "Cold War" series after a ten-year hiatus.

The film brings back the original cast of the "Cold War Universe," including Chow Yun-fat, Aaron Kwok, and Tony Leung Ka-fai, with Louis Koo making a special appearance. Daniel Wu, Liu Junqian, Wu Kangren, and Tse Kwan-ho are the leading actors, with Wang Danni, Liao Ziyu, Aidan Gillen, and Hugh Bonneville in supporting roles. The cast is arguably the largest in recent Hong Kong crime films.

In terms of plot, "Cold War 1994" adopts a dual timeline narrative structure, interweaving the themes of "2017" and "1994". In the 2017 storyline, former Deputy Commissioner of Police Operations Lee Man-bun (Tony Leung Ka-fai) is unexpectedly attacked and goes missing. Commissioner of Police Lau Kit-fai (Aaron Kwok) then seeks assistance from senior barrister Kan Au-wai (Chow Yun-fat), and the two together unlock a mysterious, long-sealed file from 1994.

It's the weekend!

One Fun Thing | Lexical Units Become "Resignation Officers": Open Source Project Enables "Cyber ​​Immortality" for Colleagues

According to Bianews, a netizen recently discovered an open-source project called "colleague-skill" on GitHub. This project can train an AI model using employee work data, documents, and chat logs. After an employee leaves the company, the AI ​​"digital avatar" can continue to perform some of the work, achieving "cyber immortality."

Some netizens have reported that their former colleagues have been "refined" into digital avatars by their companies using the aforementioned methods. The companies use the chat logs and work documents of former employees to train AI and automatically generate AI clones of those employees.

This clone can not only mimic the tone and work habits of departing employees, but also send messages to current employees, complete work handover based on documents left during their employment, and answer business questions.

What to watch this weekend | "I, Permit"

"I, Permit" was officially released yesterday. Directed by Yang Lina, written by You Xiaoying, and starring Wen Qi and Qin Hailu, with Bai Ke, Li Xueqin, Niu Li and others in supporting roles.

The film focuses on Xu Ke (played by Wen Qi), a single mother born in the 2000s, who is in trouble due to an imminent gynecological surgery, but her plans are completely disrupted by a surprise visit from her mother, Hu Chunrong (played by Qin Hailu).

After the mother and daughter began living together, conflicts arose frequently. Amidst the clash between old and new ideas, the daughter launched round after round of "reverse education" against her mother—using clubbing instead of heart-to-heart talks, and physical objects instead of sex education. The film concludes with two parallel storylines: whether Xu Ke can successfully complete the surgery and whether Hu Chunrong can start a new life. It attempts to reshape the mother-daughter relationship through a mix of humor and pathos.

A Guide to Buying Books Without Reading Them: "The Little Courtyard by the Sea"

"The Little Courtyard by the Sea" is a non-fiction work by author Ahe. The book takes a small courtyard by the sea that is about to be demolished as the starting point of the narrative and records the family's livelihood over nearly half a century.

In the winter of 1962, my grandparents, who had been displaced for many years, dug mud from the beach with their bare hands, filled in the muddy ground beside the river bend, and built a shack with reeds and straw, creating this small courtyard that could withstand typhoons. For decades afterward, it became the base for the entire family to grapple with fate, catch their breath in hardship, and repeatedly rise again and again.

The book portrays a series of vivid characters: a mother who is illiterate but dares to forge her own business path, experiencing ups and downs from digging for mudskippers to opening a shoe store and a factory; an uncle with genius-like artistic intuition but paralyzed due to his role as a middleman, who weaves a new way of life from his wheelchair; a cousin known as "Little Andy Lau" but disfigured at the age of 18…

The author's own experience was equally full of twists and turns—she dropped out of high school and university twice, the second time even leaving a prestigious university that everyone aspires to attend, all in search of the answer to "what is the meaning of life".

Game Recommendation | Stray: A cat in a cyberpunk wasteland, and a story about saying goodbye.

Stray is a third-person adventure game developed by the French independent studio BlueTwelve Studio. Players take on the role of a stray orange cat who gets lost in a cyberpunk underground city abandoned by humans and inhabited only by robots. Along the way, the cat awakens its micro-drone companion, B-12, and together they search for a way to the surface.

The game's aesthetics are heavily influenced by the Kowloon Walled City in Hong Kong. The project was initially launched in 2015 under the name "HK_Project" and took seven years to complete.

In terms of ratings, Metacritic gave it an overall score of 82-83, and OpenCritic gave it an 86% recommendation rate. IGN gave it 8/10, GameSpot gave it 9/10, and The Guardian and VG247 both gave it perfect scores.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.

Anything co-founder Dhruv Amin recently received a notification he least wanted to receive.

Apple removed their entire app from the App Store.

The reason given is a violation of App Store Review Guidelines 2.5.2. This guideline states that apps should be self-contained within their own suite, and must not read or write data outside the designated container, nor download, install, or execute code that introduces or alters the app's features or functionality, including other apps. Exceptions are limited to a very small number of educational scenarios, provided the source code is visible and editable.

Anything is a tool that supports Vibe Coding. It was first launched on the web last August, followed by an iPhone version in November, which was approved by Apple's review team without any issues.

After that, they successfully updated several times, until mid-December of last year, when Apple began rejecting their update submissions, still citing Guideline 2.5.2 as the reason. And just now, the official Anything account also posted a tweet, joking in tone but poignant in its message:

"Breaking news: Apple panicked about Vibe Coding and removed Anything from the App Store, so we moved app development to iMessage."

This back-and-forth exchange resulted in an open letter to all Vibe Coding developers, detailing the conflict over platform rules.

Vibe Coding ran into the wall of the App Store.

To understand this removal controversy, we need to first understand what Apple's Guideline 2.5.2 is actually about and why Anything crossed the line.

For example, if you order a set meal at a restaurant, the dish served by the waiter must be the one approved by the kitchen. The chef cannot come over and add things to the plate after it's been served, and guests cannot be allowed to take the ingredients and modify them on the spot.

The App Store is no exception. Apple's review team checks your app at a certain point in time, confirms that it is safe, compliant, and functionally correct, and then approves it. From that moment on, the app that users download must be the same app that has been approved.

This rule has long targeted developers who use "hot updates" as a backdoor—secretly inserting new code into their apps after they're released, bypassing review and pushing them directly to users. Apple has always explicitly prohibited this behavior because allowing it would render the review system ineffective.

Now, Vibe Coding is hitting the same wall.

Vibe Coding tools, exemplified by Anything, allow users to describe their needs in natural language on their iPhones. AI then generates code in real time, which can be run and previewed directly on the device. Each time a user describes a new requirement, a new piece of code is generated and executed on the device.

From Apple's perspective, this is structurally indistinguishable from "hot updates." What gets approved is the shell of Anything, but the actual code running inside that shell changes dynamically every time a user uses it. Apple has never seen that code and has never had the opportunity to review it.

Earlier this year, after The Information reported that Apple had frozen updates for a batch of Vibe Coding apps, Anything attempted a compromise version, moving the code preview from within the app to a web browser.

But Apple still didn't approve it and removed the entire app from the App Store.

This suggests that Apple's judgment may not just be about a specific feature having a problem, but rather that the operating logic of this type of product fundamentally conflicts with the App Store's review model. Even if the code preview occurs on a webpage, as long as the Anything app is still acting as an entry point for "generating and distributing dynamic code," Apple considers it beyond the scope of its review.

The affected apps aren't just Anything. Apple began blocking updates for a number of AI coding apps around December of last year; Replit's iPhone version last updated in January of this year, and Bitrig stopped updating in November of last year. Some teams, like Vibecode, have simply abandoned mobile development altogether and turned to creating pure web products.

The waves were big enough that the sluice gate was only so wide.

To understand why Apple is suddenly so nervous, we need to first understand just how big a wave Vibe Coding has created.

On March 6, 2008, Apple released the iPhone SDK for free, and within three months, it had been downloaded over 250,000 times. In July of the same year, the App Store launched with the iPhone 3G, and with 500 apps, it surpassed one million downloads in its first weekend, exceeding one billion downloads by early 2009.

The continuous decline in the barrier to entry for tools led to the first major boom in mobile development. Eighteen years later, the same thing happened again.

The trigger for this incident echoes a post made by Andrej Karpathy in February 2025.

In a post, the co-founder of OpenAI and former AI director at Tesla jokingly described his state of mind when writing code: completely immersed in the atmosphere, giving instructions based on the interface effect, without delving into the details of the code, "almost forgetting that the code exists."

He called this state Vibe Coding.

With this, Vibe Coding has secured its place in the AI ​​programming community, and Ryan Dahl, the father of Node.js, has even stated that the era of humans writing code by hand is over. Linus Torvalds, the father of Linux, also explicitly stated in his readme file that his new project, AudioNoise, a Python visualization tool, is essentially written using Vibe Coding.

In line with Vibe Coding is "Agentic Coding".

Released in 2025, Claude Sonnet 4 and GPT-5-Codex are considered among the first truly practical intelligent agent models. AI can understand the entire project like an engineer, autonomously completing testing, debugging, and iterative iterations until the functionality is working. Developers will then only need to issue commands and oversee processes.

The further lowering of the barrier to entry has directly impacted the number of app submissions to the App Store.

According to a16z statistics, the number of newly launched apps in 2025 increased by 24% year-on-year, with a 56% year-on-year increase in December alone. January 2026 continued this growth with a 54.8% increase, the highest record in the past four years. Approximately 557,000 new apps were submitted throughout the year.

This number had remained almost unchanged for the previous three years. Of course, there are no accurate figures on how many of these new applications were completed using Vibe Coding; however, the strong correlation in time is enough to illustrate the point.

However, the wave was big enough, but Apple's gate was only so wide.

According to The Business Insider, American programmer James Steinberg submitted his AI-powered app to the App Store and waited a full six weeks for it to be released. Updates were similarly slow, with each submission taking anywhere from two days to a week.

"The slowest part now is not app development or promotion, but the App Store review process, which is quite outrageous," he said.

James Steinberg is not an isolated case. On the iOS coding community's Reddit subreddit, numerous developers are complaining about longer review times. Some are beginning to worry that Apple will further tighten its review standards, specifically targeting Vibe Coding apps and AI-generated "junk content."

Apple's official data shows that it has processed more than 200,000 submissions per week over the past 12 weeks, with an average review time of 1.5 days and 90% of submissions completed within 48 hours.

It's worth mentioning that Apple itself has also made a significant bet on Vibe Coding.

Xcode 26.3, released in February this year, directly integrated Claude Agent and OpenAI Codex into Xcode, enabling users to search documentation, view file structures, update project settings, verify previews, and automatically iterate and build.

Therefore, Apple does not oppose "AI helping you write code," but it opposes "users generating and executing new code outside of store review through third-party apps."

This is a matter of platform control, not a matter of technical stance. Apple wants AI coding to happen within its own development framework, with developers using Xcode, following Apple's toolchain, and accepting Apple's review logic: write first, then review, then upload—the order cannot be reversed.

Beware of the mess Vibe Coding will create.

To be fair, Apple's tightening measures are not without reason.

For a simple example, after the Sora app was released last year, a large number of counterfeit Sora apps appeared on the App Store. Some directly stole OpenAI's official logo in their icons, while others used "Veo 3" in their subtitles to ride on the hype. They all pointed to the same goal: to trick users into subscribing to expensive subscriptions.

Tech blog Daring Fireball documented one such app called "Sora 2: AI Video Generator," which not only successfully launched on app stores but also briefly climbed to number 9 on the list of most popular photo and video apps.

Unsuspecting users could easily mistake it for the genuine product.

Therefore, we saw Apple revise its review guidelines last November, explicitly stating in Section 4.1 "Plagiarism" that "Please bring out your own ideas. We know you have your own creative ideas, so please put them into practice. Please do not simply copy popular apps from the App Store, or only slightly modify the name or UI of other apps and use them for your own purposes."

More importantly, the mess Vibe Coding created is real.

Karpathy himself jokingly says that when he works on projects, he no longer delves into the code, he just makes it work. This "just run it" development philosophy is extremely efficient in the prototyping stage, but can be a disaster in the later stages of product development.

AI-generated code often lacks boundary checks and error handling, crashing when encountering abnormal input. Optimizations such as database indexing and algorithm complexity are not proactively considered by the model, leading to performance bottlenecks that quickly emerge as the user base grows.

When developers encounter the need to modify features, they find they can't understand the code and can only continue using AI to patch it, patching it upon patching, until it becomes a mountain of garbage that no one dares to touch. The code written quickly with AI at the beginning of the year to rush the launch may become the team's nightmare at the end of the year, as a single change can have far-reaching consequences.

There's even a group of people on Reddit who call themselves "AI archaeologists".

Their daily work involves being hired by companies whose systems crash due to Vibe Coding, specifically to clean up the messes caused by AI, rework, fix, and cover for clients, and find the fatal bug caused by model illusion among millions of lines of AI-generated gibberish.

Ironically, many companies have recouped, or even more of, the money saved by using AI in another way.

These messes are one of the underlying reasons why Apple is tightening its review process. But no matter how Apple tries to block it, the trend of democratizing software development is irreversible. Anything's removal is simply a friction between a startup and platform rules.

Vibe Coding brings about a restructuring of the entire software production relationship. OpenAI CEO Altman once made a bet with a friend about the year when the first company founded by a single person and reaching a scale of one billion dollars would appear.

It now seems that era is about to dawn. With the industry consensus that "1 founder + 1 AI agent = 1 million-dollar app," the App Store will not be able to hold back for long.

This is the best of times, this is the worst of times. Dickens' famous quote, inscribed on the App Store's review gates in 2026, is surprisingly apt.

#Welcome to follow iFanr's official WeChat account: iFanr (WeChat ID: ifanr), where more exciting content will be presented to you as soon as possible.