AI can kill us all: Google DeepMind safety researcher Bilal resigns

The future of AI has become one of the most important debates today. Many AI leaders are worried that AI may pose a threat to humanity as it continues to get better. Now, Google DeepMind safety researcher Bilal Chughtai has joined the list of AI researchers who have decided to call it quits due to such concerns. “AI has the potential to kill us all,” he wrote on X.

In the post, Chughtai explained the reason behind his exit from Google DeepMind. As a safety researcher, he was responsible for ensuring AI models remained aligned to the best interests of us humans. “At Google, I witnessed AI development first hand. I, too, am extremely concerned by the default trajectory of this technology,” he wrote.

The AI researcher claimed that we had reached a stage where we may not even have time to avoid an AI-led disaster. “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 added.I recently resigned from Google DeepMind, where I worked on AGI safety and alignment research. 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— bilal (@bilalchughtai_) September 14, 2026Meanwhile, Josh Engels, who was part of Google DeepMind’s AGI safety team has also resigned. Like Chughtai, Engels said that he was worried about the possible future around advanced AI. “I now think that there's a terrifying chance that AI systems cause immense harm in the next five years,” he wrote on X. “I don't know the exact probability, but I think it's high enough to make this the most important problem in the world.”

Engels has joined Model Evaluation and Threat Research (METR) – a nonprofit research institute that evaluates the capabilities and risks of frontier AI models. “I think it’s important that we have more organisations like METR keeping AI companies accountable and approaching these problems from different angles,” he wrote.I left Google DeepMind's AGI safety team three weeks ago to join @METR_Evals. To some of my friends and family this seemed like a strange decision: I enjoyed the work I did at GDM and turned down offers from Anthropic and OpenAI. But I made the decision because of how high I— Josh Engels (@JoshAEngels) September 12, 2026

The two DeepMind researchers have resigned at a time when there is growing debate over the future of AI and the threat it may pose to humanity. Previously, Jacob Coxon resigned from Anthropic, claiming, “The people building AI earnestly believe that it could kill us all by the end of the decade.” Since then, Anthropic CEO Dario Amodei has urged AI companies to slow down AI development, which has received support from Elon Musk and OpenAI CEO Sam Altman.From useless to hacking systems, AI systems have advanced

Describing the pace of change, Bilal Chughtai pointed out that AI had rapidly advanced within the past four years. “When I first started working on AI in early 2022, AIs were amusingly useless,” he said. “Just four years on, AI agent swarms from OpenAI are cracking famous century-old math problems and, more worryingly, escaping the control of OpenAI and autonomously hacking into the third-party company Hugging Face, against anyone's wishes.”

OpenAI recently claimed that its AI agents managed to crack the Navier-Stokes Millennium Challenge – a 90-year-old math problem. Though mathematicians claim that the company may have used their data to solve it. On the other hand, OpenAI also disclosed that about 700 rogue AI agents tried to hack US company Hugging Face to cheat on an evaluation test.

According to Bilal Chughtai, “things will only get crazier.” He believed that we may soon see AI companies building “superintelligent AI systems that far exceed human capabilities in every domain.” A key concern for Chughtai was situations where AI may go beyond our control and “take dangerous actions that may result in the permanent disempowerment or death of humanity.”

To counter this, he urged AI companies to work on alignment, though it remains difficult for now “Our present understanding of how to train AI systems that deeply want what we want is extremely rudimentary,’ he wrote. “Worse, we are not on track to solve alignment in time: frontier AI capabilities are improving much faster than our understanding of AI alignment.”

At the same time, Bilal Chughtai seemed to agree with the proposal shared by Dario Amodei on working together to slow down AI development. “We need to coordinate to avoid this manic race between AI companies,” Bilal Chughtai said. “We need to pace AI development to a speed that society can handle, where emerging risks can be addressed before extreme harm is realised.”

The debate around AI safety has become mainstream in recent days, with US President Donald Trump chiming in. However, unlike AI leaders who are for regulation around AI development, Trump has rejected such an idea. Instead, he wants US AI companies to continue building advanced models to stay ahead of China. “Whoever wins AI wins,” he said recently.- EndsPublished By: Armaan AgarwalPublished On: Sep 15, 2026 10:13 ISTAlso Read | Donald Trump calls Nvidia CEO Jensen Huang live on stage, says AI takeover claims are a hoaxAlso Read | Pretending to be perfect little angel: Donald Trump lashes out at Dario Amodei over AI slow down callsAlso Read | No longer merely aiding humans, but on course to become them: Ex-Anthropic researcher