Former Anthropic researcher Jacob Coxon said current AI tools are safe for everyday use but warned that more powerful systems could become capable of harming humanity if industry competition outpaces safety measures. In an interview with CBS News, he argued that unrestrained models could propagate online, be misused for bioweapons, and resist shutdown, urging stricter government regulation and third‑party audits to slow the race among AI developers. Anthropic defended its safeguards, citing interpretability research and ongoing testing across cybersecurity and biology, and said it has blocked misuse attempts. The debate underscores rising pressure on U.S. policymakers and leading labs as they balance rapid innovation with mounting systemic-risk concerns.
Related article:
