AI Governance Institute
← News
Standards2026-10-25

Bills Target AI Model Theft and Political Deepfakes With Disclosure Mandates

What happened

Congress introduced two sets of bipartisan bills this week. Each targets a distinct AI governance gap, as reported in Tech bills of the week: Preventing AI model theft; Regulating AI-generated political content; and more. The first bill directs NIST AI Risk Management Framework (AI RMF 1.0) and Playbook-adjacent work at NIST. It tasks the agency with developing voluntary cybersecurity standards to defend against distillation attacks. In these attacks, adversaries send a model huge volumes of queries to copy its behavior without accessing the underlying code. It would also create an AI Cybersecurity Roundtable linking critical infrastructure operators with technology firms. A separate set of bills would require clear disclaimers on AI-generated content in political advertising and would prohibit using AI tools to interfere with voters' access to the ballot. These measures remain proposals and are not yet law. The distillation-attack threat was separately named by the NSA, CISA, and FBI as an industrial-scale model intellectual property risk in an earlier joint advisory, reinforcing the legislative concern here.

Why it matters

  • ·Organizations with proprietary AI models face a named, legislatively recognized threat: distillation attacks that copy a model's behavior through repeated querying. The NIST AI Risk Management Framework (AI RMF 1.0) and Playbook provides a basis for building protective controls now. Regulators may treat inaction as a governance gap, even before these bills pass.
  • ·Any organization involved in producing, funding, or distributing AI-assisted political communications faces potential mandatory disclosure requirements. If passed, these bills would require explicit AI-content labeling in political advertising, creating a new compliance obligation for campaigns, political action committees, media buyers, and their AI vendors.
  • ·The pairing of a cybersecurity standard-setting mandate with a content-transparency mandate in the same legislative cycle signals that Congress is treating AI governance as multi-dimensional. Compliance programs built around a single risk type, such as bias or privacy, may be structurally incomplete as federal oversight expands.

Stay ahead of stories like this

Get every US AI governance development like this one, plus the rest of the week's developments. Every Thursday.

Free every Thursday. Unsubscribe anytime.

Governance controls affected

CMP-006AI Content Watermarking and Labeling ComplianceCMP-001Multi-Jurisdiction AI Regulatory Compliance MappingSEC-005Adversarial Robustness TestingPRC-001AI Vendor Due DiligenceMGV-009AI Capability Claim Substantiation Standard

What to do now

  • ☐Ask your security or engineering team whether your proprietary AI models could be copied by an adversary sending large volumes of queries, and what controls currently prevent or detect that activity.
  • ☐Identify every product or campaign your organization is involved in that uses AI to generate or modify political advertising content, and document whether current review processes would support a mandatory disclaimer requirement.
  • ☐Map which business units produce AI-assisted content for external audiences and confirm whether your AI content labeling policy covers political communications specifically.
  • ☐Assign someone to track these bills through committee and flag any markup or hearing dates, so compliance and legal teams have lead time before any floor vote.
  • ☐Review vendor contracts for any AI tool used in content production to determine whether distillation-attack protections and content-labeling obligations are addressed in current terms.

What to watch next

If either bill advances to markup or a Senate companion is introduced, compliance teams will need to move quickly on content labeling workflows and model protection controls. Watch for NIST to publish a public comment period or draft guidance on distillation-attack standards, which would give enterprises an early benchmark even before legislation passes. Separately, state-level AI political content bills, including those already enacted in several states, may interact with any federal standard and create a patchwork compliance problem. The AI-Powered Foreign Influence Campaigns Expose Coordination Gaps Before 2026 Midterms story underscores that enforcement pressure on AI-generated political content is building from multiple directions simultaneously.

Related Coverage

Research2026-11-11

Glasswing's 94.2% Accuracy Claim Conflicts with Independent Evidence

Seven months into Anthropic's AI-driven vulnerability disclosure program, Project Glasswing, independent analysis by VulnCheck finds that only 217 of 37,966 reported findings have been marked fixed. An additional 1,654 have been flagged as duplicates. The curl open-source project rejected 19 of 34 published reports as non-vulnerabilities. This suggests a false positive rate of at least 55.9% in that sample. That figure directly conflicts with Anthropic's claimed 94.2% true positive rate. Compliance teams that have treated Anthropic's safety program metrics as evidence of vendor control quality face a due diligence gap.

Corporate Policy2026-11-14

Nvidia's Private Chip-Tracking Admission Widens Export Control Compliance Gap

Nvidia privately told US Representative Bill Foster that geolocation tracking for advanced AI chips is technically feasible, contradicting its industry association's public claim that such mandates are "potentially infeasible." The disclosure emerged in the context of the proposed Chip Security Act, which would require verified chip location data to prevent smuggling to China. Foster plans to introduce stricter follow-on legislation in 2027 that would require license verification before an advanced GPU can connect to a network.

Research2026-11-14

AI Auditor Shortage Threatens IVO Oversight Before Laws Take Effect

A growing policy consensus supports independent verification organizations (IVOs) as the primary mechanism for third-party AI oversight, but a severe shortage of qualified technical auditors threatens implementation. Proposed regimes, including Massachusetts' Mass Wins Act requiring roughly quarterly catastrophic risk evaluations and annual compliance audits, face staffing constraints at organizations such as METR and AVERI. Enterprises whose AI governance programs assume third-party audit availability face timing and coverage gaps as a result.