AI Researcher Quits Over Existential Safety Concerns

Anthropic researcher Jacob Coxon, aged 27, announced his resignation from the artificial intelligence industry yesterday, expressing profound concerns about the safety of advanced AI development. Coxon, who previously worked at OpenAI between 2023 and 2026, accused his former employers of "racing straight to self-improving superintelligence and gambling with our lives." In a widely viewed X post, Coxon warned that some AI builders genuinely believe the technology could lead to human extinction by the end of the decade. Evan Hubinger, an alignment science lead at Anthropic, corroborated Coxon's sentiment, confirming that some researchers indeed fear existential risks from advanced AI. Coxon's departure follows a pattern of AI professionals leaving frontier AI labs due to ethical and safety considerations. In 2024, multiple members of OpenAI's safety and governance teams resigned, citing a loss of confidence in the company's safety protocols. Earlier this year, Caitlin Kalinowski, an executive at OpenAI focused on robotics, resigned due to the company's contractual work with the Department of Defense. Anthropic, which has marketed itself as a more safety-focused AI developer compared to OpenAI, also experienced a significant departure. Safety researcher Mrinank Sharma left Anthropic earlier this year, citing concerns about AI risks, bioterrorism, and internal pressures to compromise on safety priorities. Stuart Russell, a British computer scientist and professor at the University of California, Berkeley, suggested that Coxon's resignation might mark a new phase of heightened awareness regarding AI risks. Russell noted that these concerns are gaining traction at a time when the public and policymakers are beginning to pay closer attention. This sentiment is further underscored by recent discussions surrounding Anthropic's Claude Mythos model and other high-profile incidents that have brought AI safety issues to the forefront of public discourse. The collective actions of these researchers highlight a growing internal dissent within leading AI organizations regarding the pace and direction of AI development, particularly concerning the potential for uncontrolled superintelligence and its existential implications for humanity. The trend indicates a critical juncture for the AI industry, prompting calls for greater transparency, robust safety measures, and a more cautious approach to developing increasingly powerful AI systems.