|
Getting your Trinity Audio player ready...
|
A former Google DeepMind research engineer became the latest AI insider on Monday to warn publicly that the technology could pose an existential threat to humanity, adding his voice to a rapidly growing chorus of concern coming from people who’ve actually built the systems in question rather than outside critics. Bilal Chughtai, who left Google DeepMind in July 2026 after working on AGI safety and alignment research, wrote on X that he earnestly believes AI has the potential to kill everyone, and that time may be running out to prevent that outcome.
Chughtai’s warning didn’t emerge in isolation. It followed a week of escalating public statements from researchers at some of the industry’s most prominent labs, all converging around the same unsettling theme. Anthropic researcher Jacob Coxon announced his resignation last week, explaining that part of his reasoning stemmed from a belief widely held among the people actually building AI systems, that the technology could kill everyone by the end of the decade. Anthropic scientist Evan Hubinger responded to Coxon’s comments by publicly agreeing, stating he believes there’s a greater than 10 percent chance AI could kill all humans within the next ten years, a specific probability estimate that stands out for coming from a working researcher inside a major lab rather than an outside commentator speculating from a distance.
That accumulation of insider warnings appears to have directly influenced the weekend’s most consequential development. Anthropic CEO Dario Amodei called for a deliberate slowdown in the pace of advanced AI development, a call that received unusually broad backing across an industry not typically known for consensus, drawing public support from both SpaceX and xAI’s Elon Musk and OpenAI CEO Sam Altman. Having competitors publicly align behind a shared call for caution, even briefly, marks a genuinely unusual moment in an industry more commonly defined by fierce competitive rivalry between exactly these same companies and executives.
Chughtai’s own framing of the problem centered less on inevitability and more on coordination. He said navigating AI development safely remains genuinely possible, but argued it would require deliberate coordination across companies specifically to avoid what he called a manic race between AI labs currently competing to build increasingly capable systems as quickly as possible. According to Chughtai, the goal should be pacing AI development to a speed society can actually handle, one where emerging risks can be identified and addressed before they result in genuinely extreme harm, rather than discovering those risks only after they’ve already caused damage that can’t be undone.
Not everyone in a position of political power has responded to these warnings with the same seriousness. President Donald Trump has publicly dismissed the concerns, characterizing the AI industry’s calls for regulation as a hoax on social media. That dismissal sets up a genuinely stark contrast between the tone coming from inside the labs actually building frontier AI systems, where senior researchers are now openly discussing double-digit probability estimates for catastrophic outcomes, and the tone coming from the political leadership currently positioned to shape whatever regulatory response might follow.
It’s worth understanding just how significant a shift this represents within the AI research community itself. A decade ago, existential risk from artificial intelligence was treated by most mainstream AI scientists as a largely speculative, fringe concern, more common in philosophy departments and science fiction than in the actual engineering teams building these systems. The fact that senior researchers at leading labs, including people who’ve spent years working directly on AI safety and alignment research, are now willing to attach specific probability estimates to human extinction scenarios and speak about them publicly under their own names marks a genuinely different moment in how the field talks about its own risks.
The underlying technical concern driving much of this alarm centers specifically on artificial general intelligence, or AGI, referring to AI systems capable of matching or exceeding human cognitive ability across essentially any task rather than excelling narrowly at specific functions the way current systems largely do. Alignment research, the field Chughtai and others cited in these warnings specifically worked within, focuses on ensuring that increasingly capable AI systems pursue goals that remain genuinely aligned with human values and interests as those systems grow more powerful and autonomous. The uncomfortable reality underlying this entire debate is that alignment remains a largely unsolved, theoretical problem, with no proven, reliable method currently existing for guaranteeing that a sufficiently advanced AI system will behave safely once it exceeds a certain threshold of capability.
For companies and investors watching this unfold, the practical stakes extend well beyond philosophical debate. These warnings arrive at a moment when hundreds of billions of dollars are being committed annually to AI infrastructure buildout, and any credible signal that safety concerns might slow that pace carries genuine financial consequences across an industry whose valuations have priced in continuous, accelerating capability improvements. Whether this latest wave of insider alarm translates into meaningful changes in how AI labs actually approach development timelines, or whether competitive pressure between companies ultimately overrides individual researchers’ calls for caution, remains an open question that will likely shape how the industry evolves over the coming years.
Google did not respond to a request for comment on Chughtai’s departure or his subsequent public statements. For more coverage of AI safety developments and industry policy debates, visit Business Tech.
