Former Google DeepMind safety researcher Bilal Chughtai warned AI could kill humanity, while another former lab researcher sees serious harm arriving within five years. Key Points: Chughtai l
Former Google DeepMind safety researcher Bilal Chughtai warned AI could kill humanity, while another former lab researcher sees serious harm arriving within five years.
- Chughtai left DeepMind in Jul. and said superintelligence could emerge within the next few years.
- He says alignment research trails capability growth and supports a coordinated slowdown, not a halt to AI development.
- Former DeepMind researcher Josh Engels sees a risk of immense harm within five years, while Anthropic and OpenAI figures have backed calls to pace development.
Chughtai, a former research engineer focused on artificial general intelligence safety and alignment, left DeepMind in Jul. and published his warning this week.
He said AI labs could build superintelligent systems within the next few years, potentially exceeding people across every domain while remaining difficult to align with human goals.
Chughtai said the central problem is alignment, writing, “I earnestly believe that AI has the potential to kill us all, and that we might be running out of time to avoid this outcome.”
He said current methods for training advanced systems to reliably pursue human objectives remain “extremely rudimentary,” while frontier capabilities are improving much faster than researchers’ understanding of alignment. He does not want AI development stopped. Instead, Chughtai wants companies to coordinate, slow the competitive race and provide greater transparency around how frontier systems are built and tested.
Also Read:CLARITY Act Faces 60-Vote Senate Test As Opposition Grows
Chughtai’s warning follows Jacob Coxon’s resignation earlier this month after three years across Anthropic and OpenAI. Anthropic alignment science lead Evan Hubinger later said he believed there was a greater than 10% chance AI could kill all humans within the next decade.
Engels, who also worked on DeepMind’s safety team, said Sept. 12 he joined evaluation nonprofit METR after rejecting Anthropic and OpenAI offers, citing a “terrifying chance” of immense AI harm within five years.
Anthropic CEO Dario Amodei has urged companies to pace capability gains, a position OpenAI CEO Sam Altman also backed as the debate widened across leading labs. The agreement matters because warnings now come from departing safety researchers as well as executives running major AI labs.
The current alarms follow a rapid shift since early 2022, when Chughtai began working on AI and described systems as far less capable than today’s frontier models. Over four years, the debate has moved from long-term safety theory toward specific concerns about autonomy, alignment and whether safety research can keep pace with capability gains.
Read Next:Bitget's API Now Reaches Gold, Forex, Indices And Crude Oil
