Holy crap, OpenAI and Anthropic are already simulating how to respond to an AI disaster Some people think something could go wrong in the next 6–12 months! Executives from AI giants like OpenAI and Anthropic are privately running simulations of how human society might react to a catastrophic AI incident. The biggest fear? AI being used to launch massive cyberattacks, crippling banking and financial systems, disrupting the internet, and even shutting down power and water supplies. Multiple insiders in the AI industry believe there’s a significant chance of a major incident happening within the next 6–12 months. OpenAI has admitted they’re conducting related emergency drills but emphasized that this doesn’t mean a disaster is certain. Anthropic, on the other hand, declined to comment. These companies are even preparing political scripts for after an incident. They’re considering how to brief U.S. lawmakers in advance about the risks, so they can influence regulatory policies in the aftermath of a major incident. Potential measures being discussed include halting the development of advanced AI models or requiring all advanced AI systems to have mandatory emergency shutdown switches. AI + quantum could disrupt the entire world—living with constant anxiety every day. Source: AXIOS