Ex DeepMind Researcher Warns AI Could Kill Us All
Left 50%
Center 0%
Right 50%
1 left · 0 center · 1 right
What happened
Bilal Chughtai, a research engineer who spent about 18 months on AGI safety and alignment at Google DeepMind in London, resigned and on September 14, 2026 posted that he "earnestly believe[s] that AI has the potential to kill us all" and that humanity "might be running out of time to avoid this outcome." He wrote that understanding of how to train systems that "deeply want what we want" is "extremely rudimentary," that frontier capabilities are improving faster than alignment, that companies might "in the next few years" build systems that "far exceed human capabilities in every domain," that he is "not confident that these AI systems will do what we want," and that development should be paced "to a speed that society can handle." He then joined BlueDot Impact, a London-based AI safety nonprofit; the comments followed similar warnings from former Anthropic and OpenAI researcher Jacob Coxon. Anthropic CEO Dario Amodei subsequently urged frontier labs "to pace" development while advancing alignment, and OpenAI CEO Sam Altman and Elon Musk expressed agreement. President Trump rejected the warnings as a "HOAX" and a "SICK conspiracy" against AI and data centers that would benefit China, saying the only needed "guardrails" are a "STRONG AND SMART (High IQ!) PRESIDENT" and that existing criminal and regulatory powers suffice, while reporting indicated the White House was considering only "limited safeguards" and industry self-regulation.
Sourcing check · Fully on the record · 2 articles
2 of 2 articles cite a named source for the story. It was on the record from the first report.
What these terms mean and how they're counted
- On the record · names a source 2 of 2
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Bilal Chughtai, Dario Amodei, Elon Musk, Sam Altman and Jacob Coxon
- Primary evidence · cites a document or data 0 of 2
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: None cited.
- Credits another outlet 0 of 2
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
- Unnamed sources 0 of 2
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Sep 15, 15:35 UTC by Business Insider, naming its source.
- Not counted 0
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out
Unpacked
Left-leaning coverage leaves out Chughtai's own claim that the technology can still be built safely. Right-leaning coverage reports that he said firms must pull back from a "manic race" and slow development to a speed society can handle. That warning, as carried by Business Insider, stops at his fear that systems will not do what we want and that alignment work is falling behind. Without that caveat, the resignation reads as a death warning rather than a demand to hit the brakes.
Right-leaning coverage leads by tying the resignation to Trump's dismissal of the threat as a "hoax." Left-leaning coverage never names the president and instead follows lab chiefs agreeing that development should slow. The White House is in one lede and absent from the other.
What did Chughtai actually see inside Google DeepMind, rather than in other labs' public mishaps, that convinced him time was running out?
Bottom line
A Business Insider reader never learns that Chughtai left with a slow-down plan, not only an extinction warning; the New York Post is the account that still has him arguing harm can be headed off.
The receipts — 3 facts one side's coverage lacks
- BlueDot Impact recently raised $25 million and is expanding to San Francisco
- Amodei told CBS that coordinating an AI slowdown will be very difficult because of military advantage
- The White House has been considering limited safeguards and leaning toward industry self-regulation
The Left View
Business Insider presents the warning as continuing a wave of insider concern, noting that while some in the industry criticized Coxon, others at Anthropic and OpenAI have echoed his fears. It highlights Chughtai's argument that alignment has "proven to be difficult," that researchers are "not on track to solve alignment in time," and that episodes such as "OpenAI's rogue agents hacking Hugging Face" "underline the risks of misaligned AI models," especially if labs build systems that surpass human intelligence. The outlet also reports BlueDot Impact said it recently raised $25 million and is expanding to San Francisco, quoting the nonprofit as "excited to welcome Bilal" to "build the institution that will make it easy for great people to gain context, upskill and get to work on AI safety in a fast and effective way."
The Right View
The New York Post casts the warning as "chilling" and sets it against Trump's dismissal of AI doomsday fears as a "HOAX," including his charge that Amodei is pretending to be a "perfect little angel" and that a "SICK conspiracy" against AI and data centers would benefit China. It quotes Coxon saying developers are "gambling with our lives" and that people building AI "earnestly believe that it could kill us all by the end of the decade," and notes Amodei's remark that a coordinated slowdown will be "very difficult because the incentives to pull ahead and the military advantage that you get from that are so large." The paper reports administration thinking that "limited safeguards" would be "the outer limits," with a lean toward industry self-regulation out of concern that "more sweeping restrictions could hand an advantage to China."
Our Take (balanced)
Left-leaning coverage's strongest argument is that insiders who worked on the systems themselves now say they "earnestly believe" AI could "kill us all," pointing to alignment as unsolved and capabilities outrunning it—a worry given weight when Amodei called for pacing and Altman and Musk agreed. Right-leaning coverage's strongest argument is that treating those warnings as a reason to hit the brakes would cede military and commercial advantage to China, that the administration regards extinction talk as a "HOAX" warranting at most "limited safeguards," and that Amodei himself said a coordinated slowdown may not be "possible" given those incentives. The central unresolved tension is whether those insider extinction claims are well-founded enough to justify slowing U.S. frontier development, or whether they are overstated relative to the strategic cost of falling behind.
2 sources
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.