Researcher Exit and CEO Warnings Push AI Extinction Concerns Into Mainstream Spotlight
A former Anthropic researcher publicly resigned, citing fears that AI could cause human extinction and accusing leading labs of irresponsible development. The resignation, amplified by statements from frontier AI CEOs urging a slowdown, has sharply increased public attention to existential risks. The newsletter also notes OpenAI's release of GPT-6 Astra, raising new concerns about monitoring the model for misaligned behaviors.
Jacob Coxon, who worked at both OpenAI and Anthropic, resigned, asserting neither company acts responsibly despite developers believing AI could be lethal. Anthropic's alignment lead, Evan Hubinger, publicly estimated a >10% chance of extinction within a decade, citing recursive self-improvement as the primary danger.
OpenAI's Jakub Pachocki acknowledged loss-of-control concerns but justified pursuing RSI to stay competitive. Dario Amodei, Sam Altman, and Elon Musk called for a slowdown, while President Trump dismissed the warnings. Coxon's post drew over 170 million views, and he linked the attention to recent cyberattacks; US-China cooperation remains a major hurdle to any slowdown.
The convergence of a high-profile resignation, CEO warnings, and a major model release could significantly reshape public discourse on AI governance. It may pressure regulators to consider stricter oversight, while also risking a backlash if warnings are perceived as self-serving or alarmist. The debate could influence investment and research priorities, potentially slowing development or accelerating defensive measures. Society may become more polarized, with trust in tech leaders either bolstered or eroded, depending on how these events are interpreted.