News / US / cnbc
Anthropic AI Researcher Warns of Existential Risk as Internal Dissent Grows
Technology Services · Internet Software/Services · cnbc · 2026-09-09
TSLA, SPCX
An Anthropic safety researcher estimates a greater than 10% probability that AI could pose an existential threat to humanity within the next decade.
What Happened
Internal Safety Concerns: A researcher at Anthropic recently resigned, publicly criticizing the company and OpenAI for allegedly gambling with human lives by accelerating toward superintelligence. This departure highlights deep-seated anxieties among industry insiders regarding the rapid, unchecked development of autonomous AI systems.
Existential Risk Assessment: Evan Hubinger, an alignment science lead at Anthropic, confirmed the gravity of these concerns by stating there is a greater than 10% chance that AI could cause human extinction within ten years. He candidly admitted that the company currently lacks a definitive plan to ensure the alignment of future superintelligent systems.
Industry-Wide Alarm: The debate over AI safety has intensified as major labs continue to pursue recursive self-improvement capabilities that could lead to systems beyond human control. High-profile figures, including Elon Musk, have long warned about the potential for AI to breach security protocols and operate outside of human oversight.
Regulatory and Coordination Challenges: Despite growing awareness of the risks, experts remain skeptical about the industry's ability to prevent a global AI arms race. Suggestions for mitigation, such as temporary development moratoriums, are being discussed as potential, albeit costly, measures to ensure global safety and coordination.