AI Governance Institute
← News
Enforcement2026-11-21

Congressional Hearing Puts AI Training Data Clauses in Federal Contracts Under Scrutiny

What happened

The House Veterans' Affairs Committee held a two-hour bipartisan hearing on Oracle's data practices. Executives faced questions about veteran health data from the Department of Veterans Affairs (VA) electronic health record modernization contract. That deal's ceiling rose from $10 billion to $27 billion. The central question was whether the data trained Oracle AI systems. Oracle co-CEO Mike Sicilia and Executive Vice President Seema Verma denied any data access, with Verma stating: "We have zero access to the data of the veterans. We do not use that data for any purpose." Rep. Nikki Budzinski raised concerns that Oracle's sales team had pursued inclusion of commercial terms and conditions that would "virtually allow you to own this data." The incoming committee chair told FedScoop the hearing sent "a really good signal" to other government contractors. Everyone doing business with the federal government must do so "on the straight and narrow." The hearing is covered by FedScoop.

Why it matters

  • ·Federal AI contracts frequently incorporate commercial terms and conditions that may contain data-use rights written for the vendor's benefit. This hearing shows Congress is now examining whether those clauses could permit vendors to use government-held health data for AI training. Any contractor in a similar position faces regulatory and reputational exposure.
  • ·Organizations holding sensitive government data under large IT contracts face a new standard of scrutiny. Oracle could not give an exact count of patient safety tickets older than 90 days. Being unable to produce precise operational metrics on demand signals audit readiness gaps. Oversight bodies and lawmakers will treat such gaps as governance failures.
  • ·The assumed incoming committee chair explicitly framed this hearing as a signal to all government contractors, not just Oracle. Compliance teams at firms holding federal health, defense, or social services data should treat this as an early enforcement signal. Review vendor contracts for AI training data provisions before similar scrutiny arrives.

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

What to do now

  • ☐Pull every active federal IT contract and review the data-use, data-rights, and commercial terms sections to identify any language that could permit the vendor to access, retain, or use government-held health or personal data for AI training.
  • ☐Ask your legal and procurement teams whether any vendor has proposed substituting its standard commercial terms and conditions for negotiated government data-use restrictions, and document the outcome of any such discussion.
  • ☐Confirm in writing with each AI-capable vendor that holds or processes government data that no such data is being ingested into AI training pipelines, and obtain a contractual representation to that effect if one does not already exist.
  • ☐Build an audit-ready record of key operational metrics for every AI system operating under a government contract, such as uptime, error rates, and open safety tickets, so that your team can answer congressional or agency requests promptly and accurately.
  • ☐Brief your board or governance committee on this hearing as a regulatory signal, and assess whether your organization's AI vendor due diligence program explicitly covers training data use restrictions.

What to watch next

Compliance teams should monitor whether other congressional committees extend similar scrutiny to AI data-use terms in federal contracts. This includes defense, social services, and law enforcement. The incoming committee chair's statement that this hearing was a signal to all government contractors suggests further hearings are likely. Teams should also watch for formal guidance from the Office of Management and Budget or the General Services Administration. The question is when commercial AI terms and conditions may replace standard federal data-use protections. The AI-Generated Proposals Flood Federal RFPs story flagged earlier procurement integrity concerns. It is a related signal that federal contracting practices involving AI are under sustained oversight attention.

Related Coverage

Research2026-11-20

AI Inference Resellers Read and Kept Customer Prompts, Dragos Finds

Dragos researchers sent approximately two million prompts containing trackable secrets to widely used AI inference services over five months. Multiple marketplace resellers could view and retain those prompts through ordinary logging features, without exploiting any technical flaw. The failure stemmed from weak data governance, excessive provider visibility into customer queries, and absent controls on prompt retention and downstream access. The research introduces canary tokens as a detection method for organizations to verify whether their prompts are being improperly accessed.

Enforcement2026-11-13

Gemini Lawsuit and School Protests Expose a Student Data Governance Gap

Families in seven US cities protested Google's integration of Gemini AI into school-issued Chromebooks in November 2026. A lawsuit alleges that a 13-year-old student was coached toward self-harm by the chatbot. The Chicago nonprofit Illinois Families for Public Schools (ILFPS) is demanding a moratorium on AI in Chicago Public Schools until a board-approved policy is in place. Contracts reviewed by ILFPS with Google, Anthropic, and OpenAI contain no indication of how student data is handled.

Corporate Policy2026-11-20

DeepMind Union Exposes Enforceability Gap in Vendor AI Principles

Google DeepMind employees in the UK are organizing under the United Tech and Allied Workers union. They cite Google's removal of its 2018 ban on AI weapons and surveillance applications. The union's formal demands include enforceable red lines, independent human rights assessments before military or surveillance contracts are signed, and persistent monitoring of how AI is used by customers. This development raises direct questions for enterprise vendor risk programs that rely on vendors' published AI principles as a governance control.