Mark Warner Unveils Four-Part AI Legislative Agenda
Sen. Mark Warner on July 21 unveiled a federal AI agenda built around six new bills and four priorities: infrastructure, competition and safety, workforce disruption, and national security. The proposals would require disclosures from large AI data centers, set standards for consumer AI agents and advanced-model testing, and fund worker retraining, but they remain proposed legislation rather than enacted rules.
Sen. Mark Warner of Virginia unveiled a package of federal AI proposals on July 21, framing six new bills and related measures as a four-part agenda for infrastructure, competition and safety, workforce change, and national security. Warner's office published the framework and his prepared Senate floor remarks; Axios and PYMNTS independently reported on the package.
The measures are proposals, not current law. Their significance lies in the breadth of systems they would touch: data centers, consumer-facing AI agents, model testing, financial-market safeguards, workforce programs, and federal procurement.
Infrastructure and consumer systems
The Data Center Tax Accountability and Disclosure Act would require large AI data centers to disclose electricity and water use, emissions, backup generation, and other operational impacts. It would also condition federal bonus depreciation on efficiency and sustainability standards.
The AI AGENT Act would establish responsibilities for consumer AI agents seeking access to major online platforms. Warner's announcement says trusted agents would have to protect sensitive information and act in users' interests, while the National Institute of Standards and Technology would develop technical standards and the Federal Trade Commission would maintain a registry.
A separate SAFE AI Act would create NIST-led testing for models' ability to generate child sexual abuse material or non-consensual intimate images, restrict federal procurement of systems that fail those safeguards, and create a private right of action for some victims. Warner also said he plans to reintroduce legislation directing financial regulators to address AI-generated content that could disrupt markets.
Workforce and model security
The proposed National Workforce Transition Fund would use revenue associated with limits on data-center tax benefits to support retraining, portable training accounts, employer redeployment programs, tuition assistance, and better workforce data. Another bill would help universities create accelerated paths into doctoral programs in AI, cybersecurity, semiconductors, quantum computing, and related fields.
For frontier models, the Secure AI Development Act would establish mandatory secure testing before deployment, update federal vulnerability-disclosure processes for AI-specific risks, improve government-industry information sharing, and create a voluntary incident-reporting system modeled on aviation safety programs.
For data and AI teams, the practical signal is policy direction rather than an immediate compliance deadline. Organizations should watch whether individual bills gain sponsors, move through committee, or change during negotiation before treating any requirement as settled.
Key Points
- 1Warner's July 21 framework groups six new bills around AI infrastructure, competition and safety, workforce disruption, and national security.
- 2The proposals cover data-center disclosures, consumer AI-agent standards, harmful-content safeguards, worker retraining, and secure testing of advanced models.
- 3None of the measures described in the package is an enacted federal requirement; implementation depends on congressional action.
Scoring Rationale
A sitting senator's six-bill framework spans infrastructure, AI agents, model security, financial markets, and workforce policy. It is broadly relevant but remains proposed legislation without enacted obligations.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
