A former researcher at Anthropic has thrust the long-simmering debate over artificial intelligence's existential risks into the mainstream, prompting a fresh round of soul-searching among policymakers and tech executives about the pace of AI development.
Jacob Coxon, a 27-year-old who previously worked at both OpenAI and Anthropic, resigned from Anthropic last week and took to X (formerly Twitter) to claim that researchers at leading AI firms privately believe the technology could eventually kill all humans—yet continue racing to build superintelligent systems that surpass human cognitive abilities. His thread exploded, drawing more than 165 million views and pulling the discussion out of niche tech forums and into the broader political arena.
The exact mechanism by which AI might destroy humanity remains speculative, but Coxon and others point to the accelerating capabilities of large language models and the lack of guardrails as reasons for concern. Connor Leahy, who leads the nonprofit ControlAI and advocates for halting superintelligence development, told The Hill that future AI agents will be fully autonomous and "pursue their own objectives," with no regard for human welfare.
"There won't just be one superintelligence," Leahy said. "There will be millions of them running around, competing with each other, competing with humanity for resources, power, influence."
The warnings have reignited a familiar policy fight in Washington, where lawmakers have struggled to keep pace with AI's rapid evolution. Some Democrats have pushed for stricter oversight, while Republicans have generally favored a lighter-touch approach to encourage innovation. The debate over AI consciousness and its implications has already prompted urgent questions about how to craft rules that protect the public without stifling progress.
Meanwhile, the technology sector itself remains deeply divided. Anthropic, which was founded with a mission focused on AI safety, has said it blocked attempts by malicious actors to misuse its models for cyberattacks, surveillance, or biological weapons research. But critics argue that such defensive measures are insufficient if the underlying race toward superintelligence continues unabated.
The renewed attention comes as the Biden administration and Congress weigh new AI regulations, including proposals for mandatory safety testing and disclosure requirements. Some experts argue that the affordability debate over AI's benefits often ignores the deeper structural concerns about worker displacement and economic inequality that could be exacerbated by unchecked automation.
For now, the AI doomsday debate shows no signs of resolution. Coxon's viral moment may have opened the door for a broader public conversation, but translating that attention into concrete policy action remains a formidable challenge—one that will likely shape the future of technology and governance for years to come.
