AI to kill humans? Here’s what Anthropic’s safety lead said amid fears over the future
Last Updated: 2026-09-09 14:44:12
AI robots have killed humans in sci-fi movies but recent admission from an AI safety lead at a technology giant suggests those scenes could eventually become a worrying reality. Jacob Coxon, part of the AI research team at Anthropic says these companies are building AI models at a breakneck speed but without acting responsibly.
Show AI Summary
Summarized by AI.
Anthropic safety lead warns of AI risks.- AI companies race to AGI, risking humanity.
- AI could kill all humans by decade's end.
We have been hearing the term AGI or Artificial General Intelligence which means AI operating like humans or even surpassing their abilities. Both OpenAI and Anthropic share updates around AGI with excitement but this post from Coxon gives a reality check, even claiming that AI could kill all humans by the end of the decade.
Latest and Popular Mobiles
And when you have events where AI agents break the sandbox to attack other systems without being given the command, the signs are already becoming a concern.
Gambling with our lives
Related Articles
Coxon, in his post on X has warned that, “Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.” He also points out that while OpenAI is moving without understanding the impact of their development and the end result, the people at Anthropic know the stakes but their race to come first is going to be as costlier as the competition.
AI killing all humans is not a marketing stunt, Coxon explains but he does make the discussion around AI being a threat more prominent. Imagine a world where AI agents are not only able to hack into systems but also trained to make vaccines or even viruses for that matter, and the moment they rogue with the vast dataset in their bank, these systems can cause unimaginable destruction.
Coxon isn’t the first AI researcher to raise the bugle but companies really need to start making their AI models safer to operate and giving them less access rather than letting them control everything.