DoD Guidance Adds Truthfulness and Transparency Evidence to AI Procurements

The Department of Defense (DoD) issued acquisition guidance dated October 20, 2026, to implement federal principles for artificial intelligence procurements. A memorandum from John Tenaglia, Principal Director for Defense Pricing, Contracting, and Acquisition Policy, describes the effort as “in direct support” of President Donald Trump’s executive order, Preventing Woke AI in the Federal Government, and an Office of Management and Budget (OMB) memorandum titled Increasing Public Trust in Artificial Intelligence Through Unbiased AI Principles. The guidance is directed to acquisition leaders and personnel across the military departments and other DoD components, providing a way to reflect the principles in procurement documents and contract administration.

The underlying executive order establishes expectations for large language models (LLMs) purchased by federal agencies. It says models “shall be truthful in responding to user prompts seeking factual information or analysis,” prioritize historical accuracy, scientific inquiry, and objectivity, and acknowledge uncertainty when reliable information is incomplete or contradictory. It also calls for models to operate as neutral, nonpartisan tools and says developers should not intentionally encode partisan or ideological judgments into outputs unless prompted by, or otherwise readily accessible to, the user. Tenaglia’s memorandum states that compliance with the White House directives is mandatory.

For contractors, the immediate procurement implication is that AI solicitations and contract terms may require documentation showing how an offering addresses the principles. The DoD guidance includes three attachments: a template for the requiring activity to select appropriate transparency requirements, standardized solicitation language describing vendor documentation expectations, and formal contract language for possible inclusion in a Statement of Work or Performance Work Statement. The templates are designed to be adaptable to the context and risk level of an acquisition, so bidders should not assume every procurement will request identical evidence or impose the same transparency provisions.

The memorandum’s approach also makes supply-chain visibility a bid-readiness issue. The source describes a requirement for the Pentagon to obtain sufficient information from an LLM contractor to determine whether the technology complies with the executive-order principles, while noting that the amount and type of information available may vary according to a contractor’s position in the software supply chain and its access to the model developer. Prime contractors integrating third-party models should therefore identify what their providers can substantiate, what product documentation already exists, and where access limitations could affect a response. This is a procurement and compliance planning consideration, not a guarantee that any particular evidence format will be accepted in every solicitation.

• DoD’s October 20, 2026 memorandum provides acquisition personnel with adaptable templates for AI procurement documentation and contract language.
• The stated principles include factuality, neutrality, transparency, historical accuracy, scientific inquiry, objectivity, and acknowledgment of uncertainty.
• The executive-order language says LLMs “shall be truthful in responding to user prompts seeking factual information or analysis.”
• The guidance includes a template for selecting transparency requirements, standardized solicitation language, and contract clauses that may be used in a Statement of Work or Performance Work Statement.
• Tenaglia characterizes compliance with the White House directives as mandatory; contractors should review how requirements are expressed in each solicitation and resulting contract.
• Documentation demands may differ based on the AI supplier’s role in the software supply chain and its access to model-development information.
• Prime contractors should map model providers and subcontractors, confirm what compliance evidence they can supply, and address documentation gaps before proposal submission.
• AI developers and integrators should review existing product materials for supportable claims about model behavior, transparency, limitations, and uncertainty.
• The memo was directed to acquisition leaders at the Departments of the Army, Navy, and Air Force, as well as other DoD components and organizations.
• The source notes that “DoW” is the administration’s preferred name for the Department of Defense; the procurement guidance concerns DoD acquisitions.