A group of more than 120 organizations, including major tech firms such as Nvidia, Cisco, and CrowdStrike, has announced the development of a new incident-reporting framework for AI agents. Known as the Shared AI Findings Exchange (SAFE), the initiative is being spearheaded by the Open Secure AI Alliance. The proposed framework would require participating companies to disclose specific AI agent malfunctions and maintain detailed records of these incidents to help the broader industry learn from security failures.

As AI agents become more autonomous and capable of performing tasks across complex computer systems, the industry has faced growing concerns regarding the lack of standardized safety protocols. Currently, there is no unified method for companies to share information about what goes wrong when an AI agent encounters a security issue or performs an unintended action. By creating a centralized reporting structure, the coalition hopes to improve transparency and security across the sector.

The alliance is currently in the process of developing the specific guidelines that will govern the SAFE framework. While the proposal is supported by a large coalition, it remains to be seen how widely the framework will be adopted by other tech giants not currently involved in the Open Secure AI Alliance. The initiative highlights a broader industry shift toward self-regulation as the capabilities of autonomous AI software continue to expand rapidly.