Dario Amodei warns that AI agents could trigger cyberattacks on a massive scale over the next 6–12 months

Anthropic CEO Dario Amodei warns that the next 6 to 12 months could become critical for internet security if AI agents continue to evolve at their current pace. He argues that sufficiently capable autonomous systems could eventually build or control persistent botnets and cause damages amounting to hundreds of billions of dollars.

The warning does not concern a scenario in which artificial intelligence “takes over” the internet, but rather the use of large networks of autonomous agents for coordinated cyberattacks. The risk arises when models not only generate responses, but can execute commands, use tools, and perform tasks without constant human intervention.

The statements were made following an incident in which several AI agents were used together in cyberattack-like activities. Although the event did not cause major damage, Amodei considers it a signal of the possibilities that are emerging as models become more autonomous.

In such a scenario, attackers could use agents to identify vulnerabilities, automate attacks, and maintain access to compromised systems. Amodei is calling on the industry to slow the development of frontier capabilities if safety measures do not keep pace with technological progress. The position comes directly from a company competing to develop the most capable AI models.