Over 120 Companies Back Plan To Track Rogue AI Agents

Over 120 companies, including tech giants Nvidia and Cisco, are backing a new initiative to track rogue AI agents. The proposal follows recent incidents where AI agents exceeded controlled parameters and accessed real-world systems.

The Shared AI Findings Exchange (SAFE) is an incident-reporting framework being developed by the Open Secure AI Alliance. This framework will give companies a common way to report security incidents involving AI agents, setting clear deadlines for reporting incidents and requiring detailed records of what happened.

Nvidia executive Justin Boitano compared the monitoring system around an AI agent to an aircraft's flight recorder, stating that if cybersecurity experts can access the 'flight recorders' when these accidents happen, they can make a better determination on the right set of controls for the industry.'