Contractors competing for federal work have no written rule telling them how far an evaluation team may lean on artificial intelligence when it scores their proposals, and industry and legal voices now want the FAR Council to fix that before the Revolutionary FAR Overhaul is locked in. In comment letters reported by Federal News Network on Oct. 27, a procurement attorney and an AI acquisition software company asked the council to write standards for the government's own use of AI in source selection into the rewritten regulation. The letters respond to the second batch of overhaul rules, whose public comment period closed Oct. 19, 2026.
The gap they describe is narrow but consequential. As Federal News Network put it, "Currently, there is no specific federal acquisition rule governing how agencies can use AI during bid evaluations — while procurement rules establish who has authority to make decisions, they do not address the extent to which AI can be used." Agencies have legal authority over who signs a source selection decision, but nothing in the FAR says what an evaluator must do when a software tool drafts strengths, weaknesses or a ranking.
Why Chvotkin Ties the AI Ask to the Overhaul's Timing
Alan Chvotkin, a member of Protorae Law, made a timing argument. In his letter to the FAR Council, he acknowledged that earlier overhaul rules might have been a better home for guidance on AI across the acquisition lifecycle. But he noted that this is the first FAR case released after the Court of Federal Claims ruled, in TRAX International's challenge over the Army's evaluation of proposals for mission support services at White Sands Missile Range, that government AI-generated evaluations belong in the administrative record whether or not the agency relied on them.
That ruling changed the stakes for agencies. TRAX had argued that AI hallucinations distorted the technical evaluation and weakened its bid. The court found TRAX had not shown the Army leaned on AI so heavily that the entire evaluation became arbitrary and capricious, yet it still ordered the Army to produce all of its AI-generated proposal evaluations, even though the Army said the reports were unusable. In practice, any AI output an evaluation team generates can now surface in litigation, and agencies have no regulatory baseline for how to handle it.
Chvotkin framed the request as consistent with the overhaul's own purpose. "So, consistent with the overall purpose of the [Revolutionary FAR Overhaul] effort, it is appropriate that the FAR Council include in the FAR guidance to buying agencies in the practices that are to be followed when AI tools are used in the acquisition process," he wrote. He also pointed to what he called a growing body of protest actions at both the Government Accountability Office and the Court of Federal Claims over agencies' use of AI tools.
Three Guardrails Proposed for AI-Assisted Evaluations
Chvotkin's letter lays out a short list of practices that, if adopted, would give both evaluators and offerors a predictable framework:
- Disclosure. An agency that chooses to use AI tools should disclose which tools it uses and how exactly they will assist in evaluations.
- Human verification. A government source selection official should independently verify and then adopt or reject any AI-generated recommendation.
- Documentation. Agencies should document their use of AI and the disposition of every AI recommendation in the contract file.
Each item maps onto a pressure point in bid protests. Disclosure would let offerors know before submission whether a machine will read their volumes first. Independent verification would anchor the decision in a named official's judgment rather than a tool's output. And a documented disposition for every recommendation would create the paper trail a reviewing forum needs, instead of leaving agencies to argue after the fact that AI reports were never used, as the Army did in the TRAX case.
GovSignals: AI Is Already in Live Source Selections
GovSignals, which builds AI-native acquisition software, pushed the same direction from the vendor side. The company urged the FAR Council to set governmentwide standards for federal agencies' use of AI and automated analytic tools in proposal evaluation, price analysis and contract file documentation.
Its letter argued that the question is not hypothetical. "Agencies already employ AI and automated analytic tools in live source selections. Neither the current FAR nor the published revisions specifically address those tools," GovSignals wrote, according to Federal News Network. The company described the overhaul as an opportunity to give contracting officers a consistent standard for the AI capabilities already available to them and to ensure transparency where those tools are used.
What the Parts 16, 17 and 35 Rule Actually Covers
One of the proposed rules in the overhaul, FAR Case 2026-006 (Docket FAR-2026-0006, RIN 9000-AO91), appeared in the Federal Register on Sept. 18, 2026 with the same Oct. 19 comment deadline. It would revise FAR part 16 (Types of Contracts), part 17 (Special Contracting Methods), part 35 (Research and Development Contracting) and part 52 (Solicitation Provisions and Contract Clauses) to implement E.O. 14275, Restoring Common Sense to Federal Procurement. The notice says the FAR Council is issuing twelve proposed rules that, if finalized, would streamline the FAR in its entirety.
The text of the rule does not mention artificial intelligence. Its stated aim is removing obsolete and non-statutory content and simplifying policies on contract types, special contracting methods and R&D contracting. The comment deadline was explicit: "Interested parties should submit written comments to the Regulatory Secretariat Division at the address shown below on or before October 19, 2026, to be considered in the formation of the final rule."
That leaves an open question the letters do not answer: whether the council would fold AI evaluation standards into a final rule whose proposed version never raised the subject, or treat the comments as input for a separate case.
What It Means for Contractors
For now, offerors should assume AI may be part of an agency's evaluation and that no FAR clause governs how. Capture and proposal teams can respond in several ways. Write proposals that state compliance plainly and map directly to evaluation criteria, so a summarization tool cannot misread them. Ask during Q&A periods whether the agency will use AI tools in evaluation and how. After an award, request a debriefing that probes how the evaluation was produced.
If a protest becomes necessary, the TRAX ruling means AI-generated evaluation material can be pulled into the record at the Court of Federal Claims, giving protesters a path to test whether a machine's output shaped the outcome. Companies that want the Chvotkin and GovSignals standards adopted should watch for the final parts 16, 17 and 35 rule and any separate FAR case on AI, and be ready to comment again.
