Secure AI Project

Secure AI Project (SAIP) is a U.S.-based public-policy organization focused on advancing legislative requirements meant to reduce severe-harm risks from advanced AI systems—especially by making large AI developers’ safety and security practices subject to transparency and verification by independent third parties. SAIP argues that while major frontier model developers have made voluntary safety/security commitments, there remain gaps: it is not clear that safety practices are “reasonable” or followed in practice, and it is difficult for outsiders to check compliance. SAIP’s core legislative push is therefore to require safety plans/protocols to be published, require incident reporting, and support mechanisms for outside review and accountability through state and federal policy. SAIP states that it does not accept corporate funding or funds from foreign governments and that its work is supported by individual donors and nonprofit institutions.