A bipartisan group of twelve US senators introduced the AI Accountability Act today, which would require companies deploying AI systems affecting more than one million users to conduct and publish annual algorithmic audits by independent third parties.
The legislation mandates testing for bias, accuracy, and safety across protected categories including race, gender, age, and disability status. Companies found in violation would face fines of up to $50,000 per day per system until compliance is achieved.
The bill also establishes a new AI Safety Office within the Department of Commerce tasked with developing technical standards and maintaining a public registry of high-impact AI systems operating in the United States.
