Resignation over AI safety concerns and statement of risk for humanity
Following a resignation at Anthropic, safety lead Evan Hubinger stated that the probability of artificial intelligence destroying humanity within a decade is more than ten percent.
Following the resignation of a safety researcher at Anthropic due to safety concerns, the company's safety lead, Evan Hubinger, emphasized that artificial intelligence could pose a fatal risk to humanity in the future.
Resignation of the Safety Researcher
Jacob Coxon, who trained artificial intelligence systems at Anthropic, announced that he was leaving his position due to the company's lax safety approach and reckless competition. Coxon, who previously trained systems for OpenAI, stated that artificial intelligence companies are risking human lives in an uncontrollable superintelligence race.
AI Race and Risks
Industry experts have long expressed concerns that self-improving artificial intelligence systems could spin out of human control. While companies accelerate their efforts to reach this goal, it is known that some of the existing code is written with the help of artificial intelligence.
Ten Percent Risk Estimation
Evan Hubinger, who leads one of Anthropic's AI safety teams, stated that self-improving AIs are progressing faster than anticipated. Hubinger said he personally estimates that the probability of AI destroying all humanity within a decade is greater than ten percent.
Lack of a Safety Plan
Hubinger admitted that Anthropic does not yet have a clear plan to ensure that advanced artificial intelligence remains safe and aligned with human values, nor is it moving in that direction. Coxon emphasized that companies are locked in a race to be first despite the risks.
Growing Concerns in the Industry
This resignation was recorded as one of the highest-profile departures at Anthropic, a company founded by former OpenAI members. In recent years, many researchers have decided to leave such companies due to similar safety concerns.