On Monday, Bilal Chughtai explained on X and LinkedIn why he left Google DeepMind. He worked there on AI safety and alignment. The sentence that made the headlines sits at the end of his paragraph: “I earnestly believe that AI has the potential to kill us all.”
What he wrote
The wording Bloomberg quotes: “At Google, I witnessed AI development first hand. I too am extremely concerned by the default trajectory of this technology. I earnestly believe that AI has the potential to kill us all.” He added that we might be running out of time to avoid that outcome.
His prescription is the one that has been circulating for a week: companies should coordinate instead of racing each other, and pace development to what society can absorb. Google did not immediately respond to Bloomberg’s request for comment.
The timing was chosen, not accidental
Chughtai left DeepMind in July. For two months he said nothing. He published on Monday — the same day Donald Trump dismissed the industry’s concerns on Truth Social as a “hoax” and a “scam”, and Beijing called the AI bosses’ warnings fear mongering.
Wait two months and then pick that particular day, and you want to be heard.
The second one this week
On September 9, Jacob Coxon quit Anthropic, accusing the leading labs of “gambling with our lives”. Evan Hubinger, an alignment lead at Anthropic, backed him publicly and put a number on it: he personally puts the odds of AI killing every human this decade above ten percent.
Coxon came from Anthropic, and from OpenAI before that. Chughtai comes from Google DeepMind. That is two of the three labs the entire debate is about, and the two statements are nearly identical.
Two cases are not a pattern, but they are not a coincidence either
You cannot read a wave into two resignations, and anyone who does is doing what the people who read an epoch into every benchmark number do. What you can say: both worked on safety, both spoke up only after leaving, and neither describes a specific incident. They describe a direction.
That is the uncomfortable part. An incident can be investigated, fixed and closed. A direction cannot. And as long as the warnings only arrive after the resignation letter, nobody knows how many people inside those buildings think the same thing and stay.