Bilal Chughtai Leaves DeepMind's AGI Safety Team
Bilal Chughtai left DeepMind’s Artificial General Intelligence (AGI) safety team in July and this week posted his reasons for departing on X.
I recently resigned from Google DeepMind (find out more), where I worked on AGI safety and alignment research. At Google, I witnessed AI development first hand. I am extremely concerned by the default trajectory of this technology. I believe AI has the potential to kill us all, and we might be running out of time to avoid this outcome.
Chughtai argues that navigating this safely requires coordination to prevent a "manic race between AI companies" and development paced "to a speed that society can handle." Google has not commented on his post.
The timing of Chughtai's resignation and subsequent warning is notable, coming during an ongoing debate about the pace of AGI development. He is the first person to publicly voice such concerns from within Google’s lab. Jacob Coxon, who recently left rival company Anthropic, made similar warnings, stating that labs were racing towards self-improving superintelligence at a perilous pace.
Chughtai's post aligns with calls for slower development pacing, as advocated by researchers like Dario Amodei. He emphasizes the need for industry-wide coordination to mitigate potential risks.
The effectiveness of these warnings remains contingent on who is listening. Some, like Emil Michael, argue that existing market forces and government engagement are sufficient. Others, such as Donald Trump, have dismissed the concerns outright. Senators Thune, Cruz, and Klobuchar are currently drafting legislation that would impose a duty of care on developers to design against catastrophic AI risks.