Jacob Coxon warns AI labs 'gambling with our lives'

The warnings are coming from people who helped build the technology. Jacob Coxon's departure follows a pattern economist Albert O. Hirschman described as 'exit' versus 'voice': leaving an institution brings the freedom to speak publicly, while staying offers access and influence. But the longer someone participates in a system they later describe as dangerous, the harder it becomes to separate principled dissent from retrospective absolution.

For those still running the frontier labs, the dilemma shifts from whether to stay to whether to slow down. Anthropic CEO Dario Amodei's essay argues for slowing the most powerful models so safety work can catch up, but the limits he proposes depend on industry-wide coordination rather than Anthropic acting alone — an acknowledgment that safety concerns run up against commercial and geopolitical competition. OpenAI CEO Sam Altman similarly argued for pacing development and said AI escaping human control is possible, suggesting leading companies are discussing collective action.

At some point, a warning has to lead to a decision about what a company will actually do differently. The outcome hinges on whether the companies racing to build the most powerful AI can agree to slow down together, and who should set the limits — a question Altman is now wrestling with publicly.