Former Google DeepMind Researcher Warns AI Could Kill All Humans

Bilal Chughtai, who left Google DeepMind in July 2026 after working on AGI safety, wrote on X that AI has the potential to kill all humans and that time to prevent it may be running short.

• Chughtai co-authored research at DeepMind and worked there as a research engineer.
• Anthropic's Evan Hubinger put the odds at greater than 10% that AI kills all humans within a decade.
• 52% of Americans say they are more concerned than excited about AI, per a Pew Research Center poll.
• 85% of Democrats and 79% of Republicans support a federal agency to oversee AI, per one survey.

Why it matters: The warnings are escalating a debate over whether the pace of AI development can be governed before serious harm occurs.

• Anthropic CEO Dario Amodei called the current approach to building advanced AI 'reckless' and called for a slowdown.
• OpenAI asked Congress on Sept. 9 to work with the company on federal guidelines for AI.

How 61 sources split on this story

Where they split: The core dispute is whether AI extinction warnings reflect genuine near-term risk or amount to, as Elon Musk put it, a 'psyop' that distracts from real-world harms and competitive pressures.

What’s next: Trump said on Sept. 14 that the only AI guardrail needed is a 'strong and smart president.'

• House Speaker Johnson called a meeting with AI executives but ruled out legislation risking a U.S. lag behind China.
• Sen. Bernie Sanders filed a bill calling for a moratorium on AI development; its prospects remain unclear.
• Can major AI labs credibly slow development without ceding ground to competitors in China or elsewhere?
• What specific safety thresholds, if any, would trigger meaningful regulatory action in Congress?