120+ Tech Companies Propose SAFE AI Incident Sharing

A coalition of more than 120 technology companies and organizations has proposed a new framework for confidentially reporting and analyzing security incidents involving artificial intelligence agents. The initiative, called the Shared AI Findings Exchange (SAFE), was detailed by the Open Secure AI Alliance during the Black Hat cybersecurity conference in Las Vegas this week.

The SAFE proposal, now open for public comment through the Linux Foundation, aims to create a standardized, blame-free channel for organizations to share details of AI operational failures, near-misses, and agent misbehavior. According to the alliance, the current lack of a community-wide framework means valuable intelligence from internal investigations often remains isolated within individual companies. The system is designed to identify recurring control failures and translate lessons into reusable defensive guidance for the broader ecosystem.

The Open Secure AI Alliance, spearheaded by NVIDIA and launched on July 27, has rapidly grown from roughly two dozen founding members to over 120 participants. Major backers include Adobe, Cisco, CrowdStrike, Hugging Face, Intel, Microsoft, Red Hat, and Visa....