When AI labs advertise safety as a competitive advantage, that claim can become a signaling device that accelerates development rather than restraining it. Multiple high-profile labs began with safety rhetoric (DeepMind, OpenAI, Anthropic) but ended up in a rivalry where being 'safer' became a spur to move faster.
— If true, regulation and oversight must target perverse incentives behind safety signaling (competition, funding, reputational markets), not just exhortations or voluntary pledges.
Yascha Mounk
2026.04.04
100% relevant
Mallaby recounts founders meeting at safety lectures and later founding rival labs that marketed themselves as 'safer,' showing safety claims operating as competitive positioning (actors: DeepMind founders, Sam Altman, Anthropic).
← Back to All Ideas