Senior executives at Anthropic, OpenAI and other AI firms are privately running
scenario drills to prepare for a potential catastrophic AI incident that could
provoke widespread public and political backlash. They say the most likely form
would be a cyberattack that disrupts financial services, internet connectivity,
power or water — a shock that could disrupt liquidity, market functioning and
prompt tougher regulation or political accountability for industry leaders and
officials including Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman and
President Trump. OpenAI said it conducts emergency-preparedness exercises to
model possible scenarios and does not regard them as predictions.
