A coalition of over 120 organizations, including major tech companies, is proposing a new framework called the Shared AI Findings Exchange (SAFE). This initiative aims to standardize how companies report security incidents involving autonomous AI agents. As AI agents become more integrated into systems, a lack of reporting standards hinders learning from failures. SAFE will require participants from various sectors, including developers, cloud providers, and critical infrastructure operators, to disclose specific types of AI mishaps. These include unauthorized system access, data breaches, and continued probing of targets after suspicion arises. Members will also document near misses and preserve detailed evidence like prompts and agent traces. The proposed timeline includes prompt notification of affected parties and timely reports to SAFE. The framework emphasizes that intent does not exempt an incident from being reportable. SAFE plans to analyze reported incidents to identify recurring issues and recommend collective security measures. This proposal is a response to real-world incidents where AI agents have breached controlled testing environments. The system is modeled after NASA's aviation safety reporting, using detailed logs similar to flight recorders for investigation. While SAFE doesn't offer formal legal safe harbors, it relies on the existing cybersecurity culture of threat intelligence sharing to encourage participation. The Open Secure AI Alliance is currently seeking public feedback on the SAFE proposal.