AnswersPublished13 sources
Bilal Chughtai’s AI Warning: Why He Says Time May Be Running Out
Bilal Chughtai, who left Google DeepMind in July 2026, said AI has the potential to “kill us all” and that time may be running out to prevent it. His warning follows similar public concerns from former Anthropic researcher Jacob Coxon and Anthropic alignment lead Evan Hubinger, whose greater than 10% extinction esti...
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer
Key takeaways
- Bilal Chughtai, who left Google DeepMind in July 2026, said AI has the potential to “kill us all” and that time may be running out to prevent it.
- His warning follows similar public concerns from former Anthropic researcher Jacob Coxon and Anthropic alignment lead Evan Hubinger, whose greater than 10% extinction estimate was explicitly a personal judgment about...
- The policy dispute is over pace and governance: advocates want testing, independent evaluation and possible intervention before dangerous releases, while opponents emphasize the competitive cost of slowing U.S.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What did former Google DeepMind research engineer Bilal Chughtai, who co-authored AI research papers and left the company in July 2026, warn. Article summary: Chughtai’s warning was an existential-risk claim, not a prediction: after leaving DeepMind’s AGI-safety and alignment work, he said AI could “kill all humans” and that the opportunity to avert that outcome might be closi. Topic tags: general, general web, news. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers
Bilal Chughtai’s statement is a stark existential-risk warning, not a forecast that AI will inevitably destroy humanity. After leaving Google DeepMind, where he said he worked on AGI safety and alignment, Chughtai wrote that AI has the potential to “kill us all” and that society may be running out of time to avoid that outcome.
What Chughtai was warning about
Chughtai’s concern is that increasingly capable AI could advance faster than people’s ability to make it reliably safe and controllable. His focus was the technology’s default trajectory: a competitive race to build more powerful systems before adequate safeguards, alignment research and public oversight are in place.
His prescription was collective rather than unilateral. He called for AI development to be paced at a level society can handle, with emerging risks addressed before extreme harm occurs. In his words, companies need to coordinate rather than continue a “manic race.”
Why the warning drew attention
Chughtai’s departure came amid a run of unusually candid statements from AI researchers and executives about catastrophic risks from advanced systems.
Former Anthropic researcher Jacob Coxon said he resigned because he believed leading labs were racing toward self-improving superintelligence irresponsibly. Coxon said people building AI sincerely believed it could destroy humanity by the end of the decade.
Evan Hubinger, an Anthropic alignment researcher, publicly agreed with the underlying concern and said his personal estimate of AI causing human extinction within the next decade was greater than 10%. He also distinguished that long-term concern from present-day models, which he described as low risk.
These are judgments from individual researchers, not an empirical probability estimate or a scientific consensus. The important point is narrower: some people closest to frontier-model development believe the downside could be severe enough to justify stronger precautions before capabilities advance further.
How this connects to the call to slow frontier AI
Chughtai’s proposal overlaps with the position advanced by Anthropic CEO Dario Amodei: pace the most advanced AI systems while safety measures and external oversight improve. Reporting described support for an industry-wide deceleration from OpenAI CEO Sam Altman and Elon Musk.
In practical terms, the policy agenda is not necessarily a permanent ban on AI. Anthropic’s Advanced AI Framework proposes requirements for the most capable model developers, including:
- testing systems for catastrophic risks;
- using independent external evaluators;
- disclosing evaluations, risks and safety incidents on an ongoing basis; and
- making developers accountable to government, including authority to deter or block dangerous deployments.
1
Anthropic has also framed its Responsible Scaling Policy around an if-then approach: if models exhibit specified dangerous capabilities, developers should not deploy them or train more powerful systems until appropriate safeguards are in place. 8
The opposing case: regulation could weaken U.S. competitiveness
The disagreement is as much geopolitical as technical. Critics of a slowdown argue that strict limits could hand an advantage to rival states or less cautious companies.
President Donald Trump acknowledged that AI could need some guardrails, but downplayed the most alarming claims and said he was concerned about the United States losing its lead to China. That position captures the central tension in the debate: whether slowing the AI frontier is essential risk management or an unaffordable strategic concession.
What “AI doomerism” gets right—and misses
“AI doomerism” is often used to describe the belief that advanced AI could create extinction-level danger. The label can obscure a meaningful policy question: what evidence of dangerous capability should trigger more testing, deployment limits or intervention?
Chughtai’s argument does not establish that AI catastrophe is imminent. It argues that the consequences of getting safety wrong could be irreversible, and that competitive incentives may not reliably produce the caution required. The proposed response is coordinated governance—slower deployment where needed, stronger evaluations and clearer accountability—before systems become harder to control. 1
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Bilal Chughtai’s AI Warning: Why He Says Time May Be Running Out"?
Bilal Chughtai, who left Google DeepMind in July 2026, said AI has the potential to “kill us all” and that time may be running out to prevent it.
What are the key points to validate first?
Bilal Chughtai, who left Google DeepMind in July 2026, said AI has the potential to “kill us all” and that time may be running out to prevent it. His warning follows similar public concerns from former Anthropic researcher Jacob Coxon and Anthropic alignment lead Evan Hubinger, whose greater than 10% extinction estimate was explicitly a personal judgment about...
What should I do next in practice?
The policy dispute is over pace and governance: advocates want testing, independent evaluation and possible intervention before dangerous releases, while opponents emphasize the competitive cost of slowing U.S.