Comment on FR Doc # 2026-12725 from Anonymous

I write as an active NSF-funded researcher to raise four issues with the draft GFA. 1. Publication costs: the draft contradicts itself The Summary of Changes for Guide 12 states that the GFA "disallows publication costs consistent with proposed revisions to 2 CFR 200." I support this. In mathematics and adjacent fields, free distribution already exists and is universally used (arXiv), and refereeing is unpaid. Article processing charges largely spend public funds on distribution the community already provides itself. Disallowing them while requiring zero-embargo deposit in NSF-PAR (Guide 21) is coherent policy. But Guide 23 states that "PAR deposits do not require the payment of any third-party fees, but those fees are still allowable expenses on NSF awards." This contradicts Guide 12. Please strike the clause "but those fees are still allowable expenses on NSF awards" so the final document states one rule: deposit is mandatory and free, and publication fees are not chargeable to awards. 2. Reconcile the burden claims with the burden additions The GFA claims to reduce administrative burden, consistent with EO 14222 and the administration's stated science policy goals. Yet it adds: a) mentoring plans for any proposal with graduate student or postdoc salary or stipend (Guide 5); b) documented IDPs for everyone meeting the new "substantial support" definition (Guide 5); c) expanded RCR training (Guide 13); d) a new Faculty Associates disclosure category (Guides 2, 14); and e) annual Foreign Financial Disclosure Reports with mandatory negative reports (Guide 18). Some of these are CHIPS and Science Act mandates. Others, including the "substantial support" threshold, the breadth of the Faculty Associates category, and the RCR expansion, are discretionary drafting choices. Please publish a net burden estimate relative to PAPPG 24-1, as the Paperwork Reduction Act requires, distinguishing statutory from discretionary additions, and justify each discretionary addition against the burden reduction goals the document invokes. If the GFA increases net compliance burden, the final document should say so rather than claim the opposite. 3. Termination and risk assessment grant broad discretion without stated criteria Guide 25 incorporates termination authority under the proposed 2 CFR 200.340, which permits termination when an award no longer effectuates agency priorities. Guide 14 reserves the right to run risk assessments "using analytical tools" with no stated criteria, no error correction process, and no opportunity for a PI to respond. Both sit poorly with the document's stated commitment to clarity and predictability. Please specify the notice and response procedures available before a priorities-based termination takes effect, and describe the criteria and contest process governing analytical risk assessments. The TRUST pilot in Guide 9 moves in the right direction; please publish its operating details. 4. The AI claim in the Summary of Changes has no corresponding text The Summary of Changes says Guide 25 "clarifies research misconduct requirements by including reference to artificial intelligence tools." Guide 25 contains no such reference. Please either strike the summary line or state that the existing fabrication, falsification, and plagiarism definition applies to AI-assisted work without modification. The existing definition already covers what matters; fabricated results are fabrication regardless of the tool. I ask NSF to confirm the standard is unchanged rather than draft a new AI-specific regime in haste.