Researcher AI quits Anthropic, warns of safety risks
ai-safety anthropic google openai
| Source: HN | Original article
An AI researcher has left Anthropic, warning that the race toward superintelligence is a dangerous gamble with human lives.
Jacob Coxon, a senior researcher who moved from OpenAI to Anthropic, has resigned, using his departure to warn that the “race to self‑improving superintelligence” is “gambling with our lives.” In a post on X, Coxon criticised both companies for what he described as “recklessly” pursuing AI systems that could outpace human control.
The resignation adds a new voice to a growing chorus of internal dissent. Earlier this month, another Anthropic researcher told the public there was “more than a 10 % chance AI could kill all humans,” and a CNN report highlighted an employee quitting over safety concerns. The pattern suggests mounting unease among those closest to the technology about the speed and opacity of development.
Why it matters is twofold. First, the public airing of safety anxieties by a senior researcher underscores the credibility gap between frontier labs and regulators, potentially fueling calls for stricter oversight. Second, the timing coincides with recent legislative moves in California, where bills backed by Anthropic and OpenAI now govern external safety evaluations, and with broader governmental scrutiny, as illustrated by AI Minister Evan Solomon’s remarks that “our No. 1 concern always is safety.”
What to watch next includes Anthropic’s official response and any internal investigations it may launch, as well as whether the resignation prompts further departures or accelerates industry‑wide safety initiatives. Regulators could also lean on the incident when shaping future policy, especially in jurisdictions already moving toward mandatory safety audits. The episode may become a catalyst for deeper debate over how, and how quickly, the sector should advance toward self‑improving AI.
Sources
Back to AIPULSEN