Overview
Severity: MEDIUM | Affected: All AI Developers | Category: policy
In a landmark move, the United States and the European Union have jointly announced the 'AI Trust & Safety Framework,' a new set of binding regulations for companies developing and deploying high-capability AI systems. The framework mandates that all designated AI developers must conduct rigorous, independent red teaming and penetration testing before public release. A key provision is the establishment of a centralized, cross-border incident reporting database, requiring companies to disclose significant AI safety failures, security breaches, and misuse incidents within 72 hours of discovery. This initiative aims to create a shared understanding of emerging threats and prevent systemic risks. Non-compliance will result in substantial fines, up to 4% of a company's global annual turnover. The framework is seen as a major step towards harmonizing international AI governance and holding developers accountable for the security and safety of their models.