AI’s existential crisis explodes but AI companies plunge ahead anyway
SiliconANGLE Robert Hof ● Covered by 85 sources
AI insiders are warning it could kill us, while the companies keep shipping new models anyway. That gap between fear and product launches is getting harder to ignore.
Based on reporting by SiliconANGLE, Robert Hof — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic researcher Jacob Coxon quit this week, saying AI labs are “gambling with our lives.” That set off another round of public panic about where this industry is headed, and it didn’t come out of nowhere. Anthropic’s alignment science lead has said there’s a better-than-10% chance AI could “kill all humans” in the next decade. OpenAI then added its own unease, with chief scientist urging a slowdown in AI research and CEO Sam Altman reportedly telling staff the company is weighing a brake on leading-edge work if others do the same — assuming regulators don’t see an antitrust problem first.
And yet the machine keeps moving. This week brought a familiar pile-up of new models, agents and apps from Meta, OpenAI, DeepSeek and others. That is the part worth sitting with: the people building the most advanced systems are also the ones sounding the alarm, but the release cycle hasn’t really noticed. The fear may be real. The pace is real too.
There’s also a more immediate constraint than philosophical dread: power and compute. A new SiliconANGLE feature argues that AI’s appetite for data centers is driven by hard physics and ugly economics, and there isn’t a quick fix. Rising local opposition to new data centers could slow expansion whether the doom talk does or not. Companies still have to persuade communities that another giant facility, and then another one, is something worth living next to.
Money, as usual, is still pouring in. Mistral AI raised almost $3.5 billion. Paris startup Arlequin AI picked up $28 million. Cognition AI landed a $32 billion Series E at a $48 billion valuation, while Positron AI raised $875 million to argue that regular RAM can help with the memory crunch in AI chips. Oracle’s cloud infrastructure revenue more than doubled and its stock rose 4% in late trading. Adobe beat expectations too, though its outlook was light enough to push shares down after hours. The market may be nervous. The AI trade is not acting like it.
My take — AI-written commentary, not fact-checked reporting
This is the oldest trick in tech: proclaim the existential hazard, then hit publish on the next product anyway. The industry has turned catastrophe into a branding exercise, and investors are still applauding from the back row. The real tell isn’t the speeches about safety; it’s that nobody is slowing down when the checks are still clearing.
Read more about this at: SiliconANGLE