State Attorneys General Demand Big Tech Records on AI Safety

A coalition of 13 state attorneys general on December 10 asked major technology firms to produce internal records detailing how they test, monitor, and remediate generative artificial intelligence chatbots and virtual assistants. The request sought documents related to model evaluation, safety protocols, automated monitoring, and the steps companies take to address harms to consumers from incorrect or fabricated outputs.

The letter, reported by Reuters, emphasized that erroneous AI content could result in violations of state law and outlined expectations for information on mitigations and ongoing oversight. Attorneys general asked for evidence of internal testing regimes, incident logs, customer complaint handling, and policies for correcting or compensating consumers harmed by misleading or false AI responses. The action reflects growing concern among state law enforcement officials about the real world consequences of generative models as they are integrated into search, customer service, and productivity tools.

Legal experts and consumer advocates have increasingly warned that generative systems can produce plausible but false statements, fabricate sources, and propagate biased or harmful content. State prosecutors have broad authority to enforce consumer protection statutes and can pursue actions when commercial products cause tangible harm. By coordinating requests, the attorneys general signaled that they intend to share information and potentially align enforcement approaches across jurisdictions.

The letter arrived amid widening scrutiny of AI from federal agencies and international authorities. U.S. regulators including the Federal Trade Commission and various congressional committees have stepped up inquiries into AI safety, transparency, and competition. Overseas, European and other national regulators have also moved to examine how powerful language models are trained, deployed, and governed. The multi state initiative mirrors that broader trend and could feed into regulatory and legislative debates at the national level.

For companies, the demands add to compliance and operational pressures. Firms developing and deploying generative AI must balance rapid product development with documentation that demonstrates safety testing, monitoring procedures, and responsive consumer remedies. The requested materials could reveal whether companies have adequate guardrails, how they track and respond to failures, and what kinds of fixes or restitutions they offer affected users.

The enforcement drive raises questions for consumers and businesses about accountability and redress. Consumers who rely on AI for medical information, financial guidance, legal summaries, or other consequential tasks may face risks if systems produce inaccurate outputs. Attorneys general are seeking to determine what companies knew about those risks and how they sought to mitigate them.

The December 10 requests are unlikely to be the last coordinated step by state enforcers. As generative AI expands into everyday services, expect continued scrutiny of testing protocols, monitoring mechanisms, and remediation practices. The inquiries may shape not only potential enforcement actions but also industry standards and public expectations about safety, transparency, and responsibility in the age of artificial intelligence.