Flooded With Chatbot Bids, Agencies Hunt for Ways to Spot AI-Written RFI Responses

Federal buyers say generative AI has flooded their market research with machine-written answers, and they are now looking for ways to separate real capability from polished chatbot text. At the SAP Public Sector Summit in Washington on Oct. 14, Transportation Department official Anil Chaudhry told the audience that response volumes have jumped roughly fiftyfold, FedScoop reported. Agencies are pushing back in two ways. They are asking industry how to spot AI-generated bids, and they are telling offerors in advance that the government may use AI of its own to review proposals.

For contractors, both changes affect the same document. A response written to clear an evaluator's first pass now has to clear a screen built to catch generic, machine-drafted text. It may also be summarized or checked by an AI tool.

From 10 Responses to 500: The Volume Problem at DOT

Chaudhry, the Transportation Department's senior advisor for AI, autonomy and national security, described the scale of the shift onstage. “We used to get maybe 10 responses to an RFI or RFP,” he said, according to FedScoop. “Now, we’re getting 500 because everybody puts it in a chatbot and sends it to us.”

Requests for information are a market research tool. Agencies use them to learn what solutions exist before they write requirements. Requests for proposals carry more weight because they lead directly to evaluation and award. Chaudhry said both channels are clogged. “Everybody in government has these issues where they’re getting overwhelmed by qualitative information,” he said. He called the volume “a whole-of-government problem,” adding that “we have whole-of-government teams working on these problems.”

Vendors have begun catering to firms that want to speed up the response process. FedScoop identified several selling AI tools for RFI drafting: InspectMind markets a system for producing first drafts of RFI responses, Nomic promotes an AI agent that drafts them, and Datagrid is also chasing the market. Tools like these cut the effort of answering a notice, which plausibly makes it easier for more firms to respond to more notices, whether or not they can do the work.

How ICE Is Asking Industry to Detect AI Slop

Immigration and Customs Enforcement turned the question back on industry. In a recent RFI quoted by FedScoop, the DHS component acknowledged the problem directly: “The Government recognizes that the proliferation of generative AI tools has made it easier to produce highly polished, technically accurate written proposals”.

The agency then asked respondents to propose countermeasures: “To ensure the Government selects a vendor with genuine, proven capabilities to execute the scope of work, what specific evaluation methods… do you recommend to differentiate genuine, demonstrated hands-on capability features creation from purely generative AI responses?”

ICE's own wording concedes that polished, accurate prose is easy to produce. ICE did not specify which methods it might adopt. The phrase “demonstrated hands-on capability” suggests evidence that is hard to generate with a chatbot, such as live demonstrations or past performance tied to named projects, though that is an inference, not something ICE stated.

Pentagon RFP Puts Offerors on Notice That AI May Read Their Bids

The government is also adopting AI for its own review. FedScoop reported that a Defense Department RFP issued earlier in 2026 disclosed possible AI use in evaluation. “The Government may employ AI as a tool to assist in the analysis and review of offeror proposals,” DOD said in accompanying documents. The listed uses included summarizing proposal content, identifying compliance with requirements, and highlighting areas of potential strength, weakness or risk.

The same documents drew a line around human authority: “Notwithstanding the use of AI tools to support the evaluation process, all final evaluation judgements and source selection decisions will be made exclusively by duly appointed Government personnel”.

Not everyone is comfortable with that arrangement. FedScoop quoted Michael Brewer of Burr and Forman's construction practice group, who wrote in an April blog post: “It is not difficult to imagine the potential for generative AI tools to hallucinate or otherwise mischaracterize proposal details that are material to an agency’s award decision”.

Why the Protest Record Now Reaches AI-Generated Evaluations

A recent Court of Federal Claims ruling shows how AI-assisted review can surface in a bid protest. In TRAX International Corp. v. United States, No. 26-796, publicly reissued Sept. 22, 2026, the court considered an Army mission support procurement at White Sands Missile Range. According to an analysis by Bradley Arant Boult Cummings attorneys, an Army analyst testing an AI platform called FAST TRACK had generated evaluations of three proposals. Some of those reports reached the contracting officer and members of the Source Selection Evaluation Board.

The Army argued those reports did not belong in the administrative record. The court held that AI evaluations the agency considered had to be included, and Robinson & Cole says it required the Army to produce them. The Bradley Arant attorneys summarized the reasoning: “An agency’s consideration of an AI evaluation is not the same thing as its reliance on that evaluation.” Once the reports were in the record, the court found passages in the Army's evaluation of one offeror that were nearly identical, sentence for sentence, to that offeror's AI-generated evaluation.

TRAX still lost. The overlap in its own evaluation and the awardee's was minimal, and TRAX did not tie any specific evaluation error to AI. Kathryn M. Rattigan of Robinson & Cole wrote that the court held “more is needed” to establish improper reliance. She argued that FAR 15.308, which requires the source selection authority to exercise “independent judgment” and document the award decision, is likely to frame future disputes. She wrote: “The central question for future protests may be less whether an agency used AI and more whether the record shows meaningful human judgment.”

What It Means for Contractors

Taken together, these developments change how firms should write and track bids:

Volume no longer gets attention. An agency receiving 500 responses where it once got 10 has a screening problem. Generic, AI-polished RFI answers are the easiest to lose in that crowd.
Proof beats prose. ICE's question suggests evaluators want verifiable, hands-on evidence. Lead with specific past-performance references and other proof that a chatbot cannot easily produce.
Read the solicitation for AI disclosures. If an RFP says the government may use AI to summarize proposals or check compliance, write clear, well-labeled compliance sections that an AI tool can parse accurately.
Ask about AI at debriefings. The Bradley Arant attorneys suggest asking whether proposal-specific AI reports were generated, when, and who received them. Under TRAX, reports that evaluators considered may need to go into the protest record.
Protests need a link to an error. Showing that an agency used AI does not win a protest by itself. The protester has to tie that use to a specific evaluation mistake or to a failure of independent human judgment.