Former Anthropic Researcher Warns AI Could Lead to Human Extinction by 2030
Jacob Coxon, a pretraining researcher at OpenAI and Anthropic, warns that rapid AI advancement poses an existential threat.
A former researcher at the AI safety firm Anthropic has issued a stark warning that the rapid, unchecked development of artificial intelligence could lead to human extinction by 2030. The alarm comes from an insider with deep technical experience in the systems that power today's leading large language models.
Jacob Coxon, who previously conducted pretraining research at both OpenAI and Anthropic, claims that the trajectory of current AI development is dangerously unstable. Coxon warns that these models could become uncontrollable, posing a feasible threat to the survival of humanity within a few years. "Technology should enhance the human experience; it should never supplant it!" Coxon stated, emphasizing the need for human-centric guardrails over raw computational power.
The X-Risk Debate
This warning arrives amid a deepening schism within the tech industry over "existential risk," or X-risk. Anthropic, the creator of the Claude LLM, has positioned itself as a safety-first organization, yet Coxon’s warnings suggest that internal safety measures may be insufficient against the speed of advancement. The debate pits optimists, who view AI as a catalyst for unprecedented human flourishing, against safety advocates who fear the emergence of a "superintelligence" capable of acting against human interests or ignoring human commands entirely.
Political and Regulatory Impact
Warnings from high-level insiders at top-tier labs lend significant weight to the argument for stricter government oversight. Coxon's claims have already gained political traction, specifically among Florida Republicans, as reported by Politico. When technical experts from the very companies building these tools sound the alarm, it provides political figures with the justification needed to push for aggressive regulation and mandatory safety standards that could slow the deployment of next-generation models.
The Path Forward
As the industry races toward Artificial General Intelligence (AGI), the focus now shifts to whether current safety frameworks can actually prevent the catastrophic scenarios Coxon describes. While the technical community remains divided on the timeline and probability of an existential event, the involvement of a researcher with experience at both OpenAI and Anthropic highlights a systemic concern across the industry's most powerful labs. Observers are now watching to see if these warnings will trigger a shift in how AI pretraining is governed or if the competitive pressure to innovate will override the calls for caution.