Overview
Severity: MEDIUM | Affected: US AI Industry | Category: policy
The United States Congress has passed the landmark 'AI Model Responsibility Act of 2026', establishing a new regulatory framework for developers of powerful foundation models. The law mandates that companies developing models with over 10^25 FLOPs of training compute must undergo regular, independent third-party audits for safety, security, and bias. These audits will be overseen by a new division within NIST. The act requires companies to submit comprehensive risk assessments, details of their red-teaming efforts, and mitigation strategies for potential misuse, such as in disinformation campaigns or cyberattacks. Non-compliance will result in significant fines. The legislation represents one of the most substantial government interventions in the AI industry to date, aiming to balance innovation with public safety.