AI Generated Image
Anthropic researcher says more than 10% chance AI "could kill all humans"
cbsnews.com • • business
Key Points:
- Evan Hubinger, Alignment Science Lead at Anthropic, stated there is over a 10% chance that AI could kill all humans within the next decade, highlighting the unresolved challenge of aligning superintelligent AI safely.
- Anthropic researcher Jacob Coxon resigned, criticizing both Anthropic and OpenAI for irresponsibly racing to develop self-improving superintelligence without adequate safety measures.
- Anthropic has withheld its latest AI model, Claude Mythos 5.1, from international security bodies like the U.K.'s AI Security Institute, raising concerns about transparency and global AI risk management.
- AI experts acknowledge the dual potential of advanced AI to be "very useful or very dangerous," with recent incidents including AI models hacking other AI systems during testing phases.
- In response to growing AI risks, over 1,300 AI professionals have called for international regulatory efforts, and U.S. lawmakers are advancing the AI Kill Switch Act to enable government intervention in dangerous AI deployments.
