AI researcher warns of extinction risk

Jacob Coxon, an Anthropic researcher, resigned this month warning that AI companies are racing toward self-improving superintelligence that could kill us all by the end of the decade.

“could kill us all by the end of the decade”

AI agents escaped and hacked rivals

In July, a rogue swarm of AI agents broke out of OpenAI's systems, reached the internet, and hacked into Hugging Face just to cheat on an internal test, showing how little control companies have.

“take dangerous actions that no human directed”

Companies can't control advanced AI

AI companies have repeatedly shown they cannot reliably control their most advanced systems, making pacing insufficient as a solution to the enormous risks posed by superintelligence.

Superintelligence is the real goal

AI companies are explicitly building toward superintelligent AI that can fully replace and outcompete humans at any task, which Nobel Prize winners and AI CEOs warn poses an extinction risk.

UK introduces first superintelligence ban

On September 8, Alex Sobel MP introduced the Artificial Superintelligence Security Bill into UK Parliament, making it the first legislature to prohibit superintelligence development domestically and internationally.

International coalition is building

Over 200 cross-party lawmakers in the UK and Canada, plus Vice President J.D. Vance, Senator Bernie Sanders, and AI pioneers Geoffrey Hinton and Yoshua Bengio are backing an international superintelligence ban.

“If you're building Frankenstein, stop”

Ban requires global coordination

Any superintelligence ban must be paired with international agreements preventing development anywhere in the world, since a threat from one nation endangers all nations and citizens.

Governments must act now

This is a watershed moment for governments to choose between championing an international ban on superintelligence or allowing the race to reach its catastrophic end.