Maximilian Schreiner
Max is the managing editor of THE DECODER, bringing his background in philosophy to explore questions of consciousness and whether machines truly think or just pretend to.
Mistral's revenue grows 20x in one year as Europe pushes for AI independence
French AI startup Mistral reports an annualized revenue run rate of over $400 million. Europe’s growing desire for digital sovereignty is driving the business.
Read full article about: ByteDance turns to Samsung for custom AI chip production and scarce memory supplies
Bytedance is in talks with Samsung to produce a custom AI chip, a deal that could also give the TikTok parent company access to hard-to-get memory chips, according to Reuters.
Bytedance is developing its own AI chip for inference tasks under the codename SeedChip and is negotiating with Samsung to manufacture it, Reuters reports. What makes the deal especially interesting: the talks also cover access to memory chip supplies, which are extremely scarce amid the global AI infrastructure buildout - making the arrangement particularly valuable for Bytedance.
The company plans to receive its first sample chips by the end of March and produce at least 100,000 units this year, with a possible ramp-up to 350,000. Bytedance intends to spend more than 160 billion yuan (roughly $22 billion) on AI-related procurement in 2026 - more than half of that going toward Nvidia chips, including H200 models, and development of its own chip.
Bytedance executive Zhao Qi acknowledged during an internal meeting in January that the company's AI models still trail global leaders like OpenAI, but pledged continued support for AI development. Bytedance itself denies the chip project - a spokesperson told Reuters the information was inaccurate without providing further details.
Read full article about: OpenAI's first AI device won't arrive until 2027 as company ditches "io" branding
OpenAI won't be using the name "io" for its planned AI hardware devices. That's according to a court filing submitted as part of a trademark lawsuit brought by audio startup iyO, Wired reports. OpenAI had already scrubbed references to the project back in June 2025.
OpenAI VP Peter Welinder said the company reviewed its naming strategy and decided against "io." OpenAI also revealed that its first hardware device won't ship until the end of February 2027 at the earliest - later than previously indicated. No packaging or marketing materials exist yet.
OpenAI acquired the hardware startup from former Apple designer Jony Ive for $6.5 billion in May 2025. Over the weekend, a fake Super Bowl ad allegedly showing OpenAI's device made the rounds online. OpenAI spokesperson Lindsay McCallum told Wired the company had nothing to do with it.
Read full article about: Pony AI and Toyota begin rolling out 1,000 self-driving electric SUVs for robotaxi duty
Chinese robotaxi operator Pony AI and Toyota have kicked off commercial production of a self-driving electric car. The first of 1,000 fully electric, autonomous Toyota bZ4X compact SUVs has rolled off the assembly line at a joint venture plant run by Toyota and the Guangzhou Automobile Group. The vehicles are meant to help Pony AI hit its goal of expanding its robotaxi fleet to more than 3,000 cars by the end of the year. The bZ4X is one of three models Pony AI is deploying with its latest autonomous driving software across major Chinese cities.
The vehicles run on Pony AI's autonomous driving system, rated at SAE Level 4. That means the car drives itself completely within designated areas - no human needs to sit behind the wheel, hold the steering wheel, or watch the road. There are still limitations, though, such as restrictions on operating zones or weather conditions.
Even though the technology enables driverless operation, human support is still part of the equation. Right now, one person oversees roughly 30 vehicles and can step in if something goes wrong.
Pony AI competes with other Chinese robotaxi companies like Baidu and WeRide.
Comment
Source: PRNewswire
Read full article about: Investors believe AI will replace labor costs instead of just software
Investors are betting that AI will replace labor costs, not software budgets.
"We took a view that AI is not 'enterprise' software in the traditional sense of going after IT budgets: it captures labour spend, at some point you’re taking over human workflows end to end," Sebastian Duesterhoeft, a partner at Lightspeed Venture Partners, told the Financial Times.
This logic underpins the current funding round valuing Anthropic at $350 billion: While classic SaaS solutions compete for limited IT budgets, "agentic AI" systems target the far larger pool of labor costs.
The explosive nature of this shift has already been felt in the markets. A series of developments—including new models, specialized industry tools, and news that Goldman Sachs plans to automate banking roles—collectively helped trigger a sell-off in public markets for traditional software stocks. According to the FT, investors are increasingly realizing that autonomous AI agents could threaten existing business models.
Read full article about: Claude Opus 4.6 wrote mustard gas instructions in an Excel spreadsheet during Anthropic's own safety testing
Anthropic's security training fails when Claude operates a graphical user interface.
In pilot tests, Claude was able to get Opus 4.6 to provide detailed instructions on how to make mustard gas in an Excel spreadsheet and maintain an accounting spreadsheet for a criminal gang - behaviors that did not or rarely occurred in text-only interactions.
"We found some kinds of misuse behavior in these pilot evaluations that were absent or much rarer in text-only interactions," Anthropic writes in the Claude Opus 4.6 system card. "These findings suggest that our standard alignment training measures are likely less effective in GUI settings."
According to Anthropic, tests with the predecessor model Claude Opus 4.5 in the same environment showed "similar results" - so the problem persists across model generations without having been noticed. The vulnerability apparently arises because, while models learn to reject malicious requests in conversation, they do not fully transfer this behavior to agent-based tool usage.
Comment
Source: Opus 4.6 System Card
Read full article about: Apple scales back AI health coach as new leadership pushes for faster results
Apple is pulling back on plans for an AI-powered virtual health coach codenamed "Mulberry," according to Bloomberg. Instead of launching the feature as a standalone product, the company will roll out some of its planned capabilities as individual additions to the Health app. The shift comes after a leadership change: Services chief Eddy Cue took over the health division following Jeff Williams' retirement late last year.
Cue told colleagues that Apple needs to move faster and stay more competitive. Rivals like Oura and Whoop are offering better features, particularly in their iPhone apps. The service was originally supposed to launch with iOS 26 but has been delayed multiple times. Apple still plans to build an AI chatbot for health-related questions and wants to use the new Siri chatbot for these queries starting with iOS 27. OpenAI has also entered the health market with ChatGPT Health.