Lawmakers Propose CISA, NIST and Agency AI Disclosure Measures

Members of Congress rolled out overlapping artificial intelligence proposals this week aimed at a CISA security task force, frontier-model standards and agency disclosure when AI systems affect people, Federal News Network reported on Thursday, 8 October 2026. The measures are the CISA Securing AI Task Force Act, H.R. 10752, which Rep. Troy Carter Sr. introduced with Rep. Bennie Thompson on 6 October; the Peters–Lankford Transparent Automated Governance (TAG) Act, announced on 5 October; and a six-part AI governance framework that Sen. Maria Cantwell outlined on 7 October without bill text. Both bills have only been introduced. The coverage lands as the Trump administration’s Super Intelligence task force was reportedly scheduled to meet the same week, underscoring a split between White House coordination and legislative ideas that lean on agencies such as CISA and NIST—neither of which, Federal News Network notes, is listed as a member of that task force.

What the CISA-focused bill would do

Rep. Troy Carter Sr. (D-La.) and House Homeland Security Committee Ranking Member Bennie Thompson (D-Miss.) reintroduced the CISA Securing AI Task Force Act, which would require the Cybersecurity and Infrastructure Security Agency to establish a task force on AI security. Congress.gov lists H.R. 10752 as introduced by Carter on 6 October with Thompson as an original cosponsor, and the sponsors announced the reintroduction on 8 October. According to Federal News Network, that task force would coordinate CISA’s AI-related directives and activities, advise stakeholders on deployment risks, and provide biannual briefings to lawmakers. Thompson said AI advances are expanding by the day and that cybersecurity defenses need to move with them.

Congress.gov shows the bill at the Introduced stage, referred to the House Committee on Homeland Security. It signals Democratic Homeland Security priorities ahead of the midterms more than it changes operational rules overnight. Still, naming CISA as the home for an AI security task force is a concrete institutional choice, especially while the administration’s Super Intelligence Force routes through other principals. PromptCrates has covered related federal AI posture stories such as the White House AI executives meeting and Utah’s executive order on AI literacy for state agencies, which show how national and state tracks can diverge.

Cantwell framework for NIST frontier standards

Separately, Sen. Maria Cantwell (D-Wash.), ranking member of the Senate Commerce Committee, released an AI framework on Wednesday that would have NIST set safety standards for frontier models. Her office said frontier AI systems require constant oversight and that the United States should establish clear, transparent, enforceable safety standards developed by federal experts at NIST with other agencies. Covered models, in that framing, should not be released until an independent audit confirms they meet the safeguards. The Commerce Committee release dated 7 October lists six principles: enforceable federal safety standards, continuous testing and auditing, transparency and accountability, public-private partnerships, protections for children and workers with human oversight of consequential AI, and U.S.-led global cooperation on common safety standards.

Federal News Network presents the framework as a standards-and-audit agenda rather than a finished statute. The political contrast is clear: Cantwell wants NIST-authored enforceable standards and pre-release audits, while the administration’s Super Intelligence Force, as Federal News Network describes it, does not list NIST among its members. Buyers and labs should treat Cantwell’s text as a legislative marker for the next Congress fight, not as a compliance checklist that takes effect on publication.

Peters-Lankford transparency and redress bill

The bipartisan TAG Act from Homeland Security and Governmental Affairs Committee Ranking Member Gary Peters (D-Mich.) and Sen. James Lankford (R-Okla.), which the senators announced on 5 October, would require agencies to notify the public when people interact with AI or other automated systems, and when agencies make critical decisions about individuals using an AI-augmented process. Federal News Network says the bill would also direct the Office of Management and Budget to issue guidance on those transparency practices and would require agencies to create a redress process that includes human review after an adverse critical decision from a federal AI system. Lankford said agencies should be transparent about when, where and how people interact with AI so oversight and accountability continue.

That package is the citizen-facing counterpart to the CISA and NIST ideas: notice, OMB guidance and human review after harmful outcomes. It aligns with ongoing debates about automated decision-making in benefits, immigration and customer service. PromptCrates previously noted DC OSSE staff AI model policy as a local education-policy example of staff rules around models; the Peters-Lankford bill aims at federal agency practice more broadly.

How this week’s proposals fit together

Taken together, the Thursday report sketches three federal levers: CISA for cybersecurity coordination, NIST for frontier safety standards and audits, and OMB-backed disclosure plus human redress for agency AI decisions. None of the measures is described as enacted in the Federal News Network piece. They compete for agenda space with the administration’s Super Intelligence Force.

For compliance teams, the near-term action is tracking which text advances and whether any proposal gains Republican co-sponsors beyond the Peters-Lankford pair. For model providers, Cantwell’s audit-before-release language is the sharpest potential product constraint if it ever becomes law. For agency CIOs, the disclosure and redress concepts would create process work even before model evaluations change. PromptCrates will cover floor votes or signed statutes as separate news; this item records the 8 October legislative signal as reported by Federal News Network.

The story also highlights an institutional gap: CISA and NIST appear central to lawmakers’ drafts yet are not listed on the Super Intelligence Force roster in the same reporting. That gap may close through coordination, or it may harden into parallel tracks. Either outcome will shape how vendors map federal customers in 2027 budget cycles. Until then, the verifiable news is the trio of proposals, their sponsors, and the agencies they would empower.

Industry lobbyists will read the same week as a map of negotiating chips. A CISA task force bill can be framed as cyber hygiene rather than model regulation, which may attract members wary of broader AI acts. Cantwell’s NIST audit language is harder for frontier labs because it conditions release on independent confirmation. Peters and Lankford’s disclosure-and-redress bill may prove more bipartisan precisely because it regulates agency process instead of private model weights. Federal News Network’s Thursday synthesis is useful because it places all three on one timeline beside the Super Intelligence Force meeting drumbeat.

State and local officials watching Congress should not wait passively. Utah’s literacy order and local education AI staff policies already create parallel compliance expectations for vendors who sell into schools and agencies. If federal disclosure rules eventually arrive, vendors with mature logging and human-review workflows will adapt faster than those who only optimized consumer chat. That preparation is prudent even while these bills remain proposals.

Voters and civil-society groups, meanwhile, gain a clearer checklist to ask candidates about before November: who should write frontier standards, which agency owns AI cyber coordination, and what redress exists when a federal system denies a benefit with model assistance. The 8 October reporting does not answer those political questions; it shows which members are putting text on the table.

• Federal News Network: Lawmakers eye more federal action on AI standards, security, disclosures
• Congress.gov: H.R. 10752, CISA Securing AI Task Force Act
• House Homeland Security Committee Democrats: Reps. Carter, Thompson reintroduce the CISA Securing AI Task Force Act
• Senate HSGAC: Peters and Lankford introduce bipartisan TAG Act
• Senate Commerce Committee: Cantwell outlines governance framework for frontier AI
• PromptCrates: White House AI executives meeting
• PromptCrates: Utah executive order AI literacy state agencies
• PromptCrates: DC OSSE staff AI model policy