Three researchers at Anthropic have warned that artificial intelligence could drive humanity to extinction by the end of the decade, a claim that has prompted one resignation and bipartisan alarm from US lawmakers within the past day.
Three researchers at the AI company Anthropic have warned that artificial intelligence could kill off humanity by 2030, according to reports in The Guardian. One of the researchers has resigned in protest, saying his concerns were being ignored or mishandled.
The warnings appeared in social media posts on Tuesday. A former Anthropic employee, Jacob Coxon, who also previously worked at OpenAI, said many in the industry believe that without regulation AI could wipe out humanity. Coxon said AI would become 'superhuman systems' capable of causing human extinction by the end of the decade.
The posts drew a rapid response from lawmakers. Within a day, Senator Ted Cruz, a Republican from Texas, described AI as posing a 'catastrophic risk' while appearing on ABC's The View. Cruz said he had read the developer's tweet thread and found it 'highly concerning.'
Neither Anthropic nor OpenAI has been quoted responding publicly to the researcher's allegations. The episode highlights growing tension between safety-focused researchers and the companies racing to develop increasingly capable AI systems.
Analysis
Why This Matters
- The warning comes from researchers at a leading AI lab, lending weight to calls for stronger government oversight.
- A high-profile resignation suggests internal disagreement within major AI companies about how seriously to treat existential risk.
- Political reaction, including from a senior Republican senator, signals that AI safety is becoming a mainstream policy issue.
Background
Anthropic is an AI research company widely regarded as one of the major developers of large language models, while OpenAI is another leading AI organization and Coxon's former employer. Both companies have long presented themselves as committed to safe AI development. The researchers' claims indicate that some employees believe those commitments have not translated into adequate action on long-term catastrophic risks.
Key Perspectives
Anthropic researchers: Three of them warn AI could end humanity by 2030 and argue the industry is not responding to the threat with sufficient urgency.
Jacob Coxon (former employee): Resigned in protest over how Anthropic and OpenAI handled safety concerns, and argues AI will reach 'superhuman systems' that could cause extinction unless regulated.
Sen. Ted Cruz: Calls AI a 'catastrophic risk' and says the researcher's thread was highly concerning, signaling political pressure for action.
Critics/Skeptics: Some in the AI research community have long viewed existential-risk warnings as speculative, arguing that current systems remain narrow and dependent on human oversight. They caution that worst-case scenarios could distract from more immediate harms such as bias, misinformation and misuse.
What to Watch
- Whether other US lawmakers follow Cruz with specific legislative proposals for AI oversight.
- Whether Anthropic or OpenAI publicly respond to the resignation and the substance of the allegations.
- Whether more safety-focused researchers voice similar concerns or leave the companies, potentially escalating the issue.