Technology
Anthropic researcher warns of severe AI catastrophe risks after colleague resigns
CNBC reports that an Anthropic lead warned AI risks could prove fatal after a researcher stepped down.
The short version
- An Anthropic safety researcher stepped down, publicly alleging that major AI developers are recklessly pursuing superintelligence.[CNBC]
- Following the departure, an alignment science lead at the company estimated a greater than 10% risk of AI killing all humans within ten years.[CNBC]
- The alignment lead acknowledged that Anthropic currently lacks an established plan to solve safety alignment for superintelligent systems.[CNBC]
Key facts
- An Anthropic safety researcher resigned, alleging that AI developers are gambling with lives in a race toward self-improving superintelligence.[CNBC]
- An alignment science lead at Anthropic stated that there is more than a 10% probability AI could kill all humans within the next decade.[CNBC]
- The alignment science lead noted that Anthropic does not yet possess a plan to solve alignment for superintelligence.[CNBC]
- Recursive self-improvement remains technically impossible at present, though AI laboratories are actively pursuing it.[CNBC]
What remains uncertain
- Whether Anthropic or other AI labs can formulate a functional alignment plan before advanced recursive self-improvement capabilities emerge.[CNBC]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.