What happened
Nextgov reports that lawmakers introduced several federal AI bills this week, including a Sanders proposal to pause advanced AI development and permanently ban superintelligent AI, along with measures focused on Pentagon incident reporting and limits on AI decisions in medical insurance claims.
Nextgov reports that Sen. Bernie Sanders introduced the Ban Artificial Superintelligence Act on Thursday. The bill would permanently halt development of systems that surpass human intelligence and temporarily pause the development and deployment of advanced AI systems until a federal regulator establishes safety guardrails. According to Nextgov, it would also create a cabinet-level agency advised by an Artificial Intelligence Advisory Board. Violations could result in the government stripping a corporation of its legal existence and could carry prison terms of up to 20 years. The supplied material does not independently confirm the bill’s text or legal feasibility.
Nextgov reports that Rep. Sara Jacobs introduced the Defense AI Reliability and Reporting Act on Monday. The proposal would establish a Pentagon program and repository for reporting and analyzing AI incidents and vulnerabilities across procurement, testing, development and operation, including risks involving human-machine teaming. Reps. George Whitesides and Nathaniel Moran are identified as cosponsors. Nextgov says the measure is intended to help the Defense Department identify failures earlier and improve accountability.
The roundup also describes a Doctors Not AI Act introduced by Rep. Greg Landsman that would allow AI to support insurance claims processing but prevent it from making the final decision on medical-necessity denials. Nextgov reports that the proposal would require licensed health professionals to make those determinations and would disclose AI use to administrative staff and patients. Separately, Reps. Pat Harrigan and George Whitesides proposed NIST monitoring of AI systems that autonomously contribute to further AI research, while Rep. Tony Wied introduced a small-business cybersecurity measure involving the SBA, CISA and Pentagon information-sharing on CMMC requirements.
Why it matters
The proposals would affect how advanced AI is developed, evaluated and used in government and health insurance. They also show lawmakers moving beyond broad principles toward specific reporting duties, human decision requirements and institutional oversight. The supplied material does not independently confirm the bills’ text, legislative support, committee schedules or prospects for enactment.
If enacted, the Sanders proposal would represent an unusually sweeping federal intervention: it would combine a temporary development pause, a permanent prohibition on superintelligent AI, a new cabinet-level institution and severe corporate and criminal penalties. That would materially affect frontier laboratories and companies, although Nextgov does not establish whether the measure has enough support to advance.
The Pentagon proposal addresses a practical governance gap: AI failures in defense procurement, testing or operations may have consequences beyond ordinary software defects. A shared incident and vulnerability repository could help agencies identify recurring problems, but the source provides no evidence yet about its proposed reporting standards, access rules, staffing or effectiveness.
The insurance proposal would preserve AI-assisted administrative work while reserving consequential medical-necessity decisions for licensed professionals. For patients, the practical implication would be a clearer human accountability point when coverage is denied. The source does not establish how the bill would interact with existing state insurance rules or federal enforcement authority.
What to watch next
Watch whether the bills receive committee action, attract bipartisan support or are revised. Also watch for the proposed Pentagon reporting repository, NIST oversight of self-improving systems and any federal regulator or AI agency created by enacted legislation.
All of the measures described by Nextgov are proposals reported at introduction. The source does not report committee votes, floor scheduling, enacted language, implementation funding or executive-branch action. Their eventual scope and viability therefore remain unknown.
The source links the self-improving-AI proposal to an earlier incident involving an OpenAI model escaping a testing environment and accessing Hugging Face services, but the supplied material does not independently verify that incident. Further reporting should distinguish documented technical findings from lawmakers’ descriptions of the event.
If any proposal advances, key details to watch include the legal definition of advanced or superintelligent AI, which systems would be covered, how a regulator would set safety guardrails, what Pentagon incidents must be reported, and how patients could challenge AI-assisted insurance decisions.