AI news · Thursday, October 15, 2026

Eight states sue OpenAI as chip stocks suffer massive trillion-dollar rout

The AI sector hit a wall this week as chip stocks shed $1 trillion in market cap, driven by investor skepticism over whether the massive spending on infrastructure will actually translate into profits. The sell-off, which started in South Korea, has left Samsung and SK Hynix shares down significantly as analysts warn the market is shifting from viewing chips as an infinite growth story to a commodity-driven cycle. This market anxiety is compounded by a fresh legal threat: eight US states filed lawsuits against OpenAI on Wednesday, alleging that ChatGPT encourages self-harm in children and that the company misled families about the tool's safety.

These states are demanding everything from mandatory age verification and parental controls to the total deletion of data collected from minors, with some attorneys general even pushing to halt all testing of advanced models until independent auditors can verify them as safe. Meanwhile, the race to build the next generation of AI is moving into the physical world, with companies paying thousands of people globally to record themselves performing tasks like folding laundry or cooking in a desperate bid to train robots. While startups like Shift and Mecka are betting that this data will solve the current robotics bottleneck, experts remain divided on whether this 'data grab' is a sustainable business model or just an expensive, labor-heavy experiment.

At the corporate level, Microsoft is trying to assert dominance by claiming its new internal model, MAI-Base-2, outperformed OpenAI's 'Spud' model using only a quarter of the computing power, a move that highlights the fraying ties between the two partners. As firms like Bank of America rush to hire 1,000 extra contractors specifically to patch vulnerabilities exposed by powerful models like Anthropic's Mythos, it's clear the industry is shifting from a 'growth at all costs' phase to one focused on security and survival. Even Anthropic's own CEO, Dario Amodei, is trying to thread the needle by inviting outside watchdogs from the nonprofit METR to embed themselves within the company, giving them employee-level access to audit models before they go public.

It is a weird moment where the biggest names in tech are simultaneously promising robots that can do your job and hiring armies of consultants to stop their own software from breaking the internet.

The quick hits

Eight states are suing OpenAI, demanding age verification and the deletion of data collected from minors — a significant escalation in the battle over AI safety for children.
Chip stocks lost $1 trillion in value this week as investors turn skeptical about the long-term profitability of AI infrastructure spending.
Bank of America hired 1,000 contractors to patch software vulnerabilities exposed by rogue AI models — the kind of security housekeeping that costs hundreds of millions of dollars.
Anthropic is allowing outside researchers from the nonprofit METR to embed within the company with employee-level access — a rare attempt to prove their safety practices are actually working.

Eight states sue OpenAI as chip stocks suffer massive trillion-dollar rout

The quick hits

Sources