Bayeslord Warns Mandatory AI Safety Testing Risks Abuse

Pseudonymous e/acc co-founder bayeslord agrees with Dario Amodei's call for AI oversight yet cautions that mandatory safety testing will likely be exploited by powerful interests. He notes the absence of any market or governance solution that simultaneously prevents abuse and addresses externalities such as bio risks. Anthropic researcher Sholto Douglas acknowledges the difficulties of preventing power abuse but still supports some form of testing. The exchange highlights ongoing tension between the recognized need for safeguards and skepticism toward regulatory mechanisms.