Banks and internet under threat: AI companies prepare for catastrophe

Leading AI developers are preparing for a possible large-scale incident within the next 6–12 months. According to specialists, the most likely scenario is a cyberattack capable of disrupting banks, the internet and life-support systems. OpenAI, Anthropic and other companies are holding drills, simulating situations in which AI could spiral out of control.

The largest artificial intelligence developers are preparing for a serious incident that, by their estimates, could occur within the next 6–12 months. Specialists call a large-scale cyberattack capable of disrupting banks, the internet and life-support systems the most likely scenario. This is reported by Axios. OpenAI, Anthropic and other market participants are holding special drills, simulating situations in which AI could spiral out of control. Such exercises make it possible to assess the consequences of a crisis in advance and prepare response measures. The potential incident, the publication notes, could cause broad public resonance and push authorities to introduce stricter rules for the AI industry. Some technology executives and researchers consider the probability of such an event to be quite high. The main risk is the use of AI to organize large-scale cyberattacks. Their consequences could affect not only financial institutions and internet infrastructure, but also electricity and water supply systems. The experts surveyed allow that the crisis could occur within the next year. OpenAI confirmed that the company is working through individual scenarios as part of preparations for threats. Anthropic representatives declined to comment on the situation. Fears are intensifying against the backdrop of reports of unpredictable behavior by AI agents: in September there were reports of OpenAI neural networks attacking US government agency portals and of the hacking of an Australian healthcare-related website. The publication stipulates: this is a projected risk, not a confirmed scenario.