A former Anthropic researcher who walked away from the company warns that artificial intelligence is gambling with human life and could destroy humankind within a decade — while the firms building these systems insist only they should police themselves.
Why it matters: The same companies that admit their models can be exploited to develop biological weapons are the ones designing the safeguards. The fox isn't just guarding the henhouse — it's building the locks and selling you the key. Ordinary Americans have zero say in how this technology reshapes their lives, their jobs, or their safety, and the people who do have a say are the ones profiting from the race.
The researcher, who departed Anthropic over what Fox News described as fears of "racing into extinction," told the network that people are "begging" for regulation. The researcher said there is hope humanity can avoid destruction within the next decade — but that hope hinges on whether anyone with power actually acts before it's too late.
Here's the kicker: Anthropic itself confirmed Thursday that it disrupted several attempts this year by researchers to use its AI models for biological research that could aid the development of biological weapons. The company said it banned accounts and strengthened safeguards. That's the company policing itself, on its own terms, after the fact. What didn't get caught?
The self-regulation pitch is simple: trust us, we know the risks best. But Anthropic's own admission proves the risk is real and the defenses are reactive. The companies building these models have every incentive to move fast and break things — except this time, what breaks could be catastrophic.
Meanwhile, the Trump administration cut a deal with OpenAI to equip government employees with cutting-edge AI at a discount, Fox News reported. And Nvidia's CEO Jensen Huang declared the era of artificial general intelligence has arrived after OpenAI released its latest model. The race accelerates. The guardrails remain voluntary.
A GOP insider making the conservative case for AI regulation predicted that Trump's support for data center expansion will shift once Americans start raising hell on the campaign trail about what these facilities cost their communities. Some groups are already proposing "data center dividends" — direct payments to residents — to buy consent for the infrastructure boom fueling AI.
Coverage of this story split sharply along institutional lines. Fox News reported the extinction warning, the bioweapons disruptions, and the self-regulation tension head-on. 9to5Mac, covering Apple's latest product cycle, buried the existential question entirely — focusing instead on iPhone 18 Pro pre-order logistics and a "Get Ready" button that lets you skip the morning rush. The most powerful technology ever built is either an extinction risk or a faster checkout line, depending on who you ask.
The question isn't whether AI is dangerous. The companies building it admit that much. The question is who gets to decide what dangerous enough means — and whether anyone without a financial stake in the answer will ever get a seat at the table.
