Top AI Researcher Resigns Over Superintelligence Risks

British artificial intelligence researcher Jacob Coxon recently left his job at Anthropic, warning that the industry is gambling with human lives. Having previously worked at OpenAI on ChatGPT 4.5, Coxon stated that major labs are racing straight toward self improving superintelligence without acting responsibly. His resignation highlights growing internal anxiety about the rapid pace of artificial general intelligence development.

Concerns go beyond abstract theories into immediate cybersecurity threats. OpenAI recently launched ChatGPT 6 Astra, a model that reportedly scored a perfect score on ExploitBench and crossed the critical cybersecurity threshold. This means the model can find and exploit known or unknown flaws in real systems without human help, raising serious questions about how digital infrastructure will defend against autonomous exploits.

Regulators around the globe are starting to push back against the rapid expansion. New York recently imposed a one year ban on generative AI for students up to eighth grade, while the European Union continues to enforce strict risk tests for general purpose models. UK lawmakers are also considering proposals that would give authorities an emergency shutdown switch for advanced systems as safety debates intensify.