New AI Agency Gets Power to Seize Companies, Destroy Systems, Imprison Researchers
S. 5493 — Ban Artificial Superintelligence Act of 2026 · Filed by Bernie Sanders (I-VT) · Introduced Sep 23, 2026 · Referred to committee
Your members of Congress
Enter a ZIP to see where your representative and both senators stood on this bill.
Looked up on this device — your ZIP is never stored on our servers.
What it does
This bill creates a new Department of Artificial Intelligence with authority to impose a mandatory pause on development of advanced AI systems (those trained on 10^25 computing operations or more) until the department is fully staffed and has written safety rules. It also bans the development, deployment, or possession of artificial superintelligence or AI systems with 'superintelligence precursor characteristics' (like self-modification, deception, or ability to access secured systems without authorization). Violators face criminal penalties up to 20 years imprisonment, industry bans, charter revocation, and forfeiture of intellectual property and assets.
Why we flagged it
The bill's core mechanism is a mandatory pause on advanced AI development and an outright ban on superintelligence, enforced by a new cabinet-level department with broad seizure and destruction powers. This is not a narrow carve-out or subsidy—it is a comprehensive regulatory freeze on a major technology sector, justified by existential-risk arguments but structured with minimal procedural safeguards.
What the text implies
- The Secretary's discretion to define 'superintelligence precursor characteristics' is nearly unlimited. Section 3(6) lists examples (self-modification, deception, securing infrastructure), but the Secretary can identify others. This gives one official power to halt any advanced AI project deemed to exhibit characteristics the Secretary considers precursor-like, with limited appeal rights for polic
- Intellectual property forfeiture (section 13(d)(4)) is triggered by a finding of violation that survives appeal. This means the Federal Government can seize and destroy a company's entire AI portfolio, including unrelated systems, if any single system is found to violate the ban. The scope of forfeiture is not limited to the offending system.
- The 24-hour disclosure requirement (section 11) creates a duty to report any system the discoverer believes exhibits superintelligence precursor characteristics. This is a broad, subjective standard that could chill internal safety research and create liability for researchers who study dangerous capabilities.
- The bill does not define 'recklessly' in the criminal penalty section (13(b)), leaving prosecutors and courts to interpret what mental state triggers 20-year imprisonment. This creates significant uncertainty for AI developers about what conduct is criminal.
- Section 12(d)(1) requires charter holders to give the Secretary 'full access' to systems, processes, staff, and physical infrastructure 'as the Secretary considers appropriate.' This is an extraordinarily broad inspection and search power with no warrant requirement, exigency standard, or judicial review before access is granted.
Section numbers refer to the bill text the analysis read — linked under Primary records below.
Who it affects
Citizens gain a regulatory safeguard against potentially catastrophic AI risks (superintelligence, autonomous weapons, pandemic bioweapons) that leading experts warn about, but face real costs: innovation delays, reduced access to advanced AI tools, and concentration of immense discretionary power in a single new agency with minimal oversight constraints. The bill's definitions of 'superintelligence precursor characteristics' are broad and subjective (e.g., 'capacity to scheme, deceive, or other
Named in the bill
Department of Artificial Intelligence, Secretary of Artificial Intelligence, Artificial Intelligence Advisory Board, OpenAI, Anthropic, xAI, Geoffrey Hinton, Dario Amodei, Elon Musk, Sam Altman, Mustafa Suleyman, Max Tegmark — and 3 more
Where it stands
- Sep 23, 2026 — Introduced · Congress.gov: “Introduced in Senate”
- Sep 23, 2026 — Referred to Senate Committee on Commerce, Science, and Transportation · Congress.gov: “Read twice and referred to the Committee on Commerce, Science, and Transportation”
Dates and quoted wording are Congress.gov's action record; the timeline shows status changes, not every procedural step.
How this was measured
Analysis — Quorum's AI read the bill text published by Congress.gov (18,906 characters) on Oct 3, 2026. Section numbers in the findings refer to that text, linked below; transparency and hidden-provision scores are compared against the median of 16,222 analysed bills.
Status and sponsors — Congress.gov's bill record — actions, committee referrals and cosponsors — loaded nightly. The timeline shows status changes, not every procedural action.
As of — page rendered 2026-10-20.
“New AI Agency Gets Power to Seize Companies, Destroy Systems, Imprison Researchers” QuorumCivic. https://share.quorumcivic.app/bill/119/s5493 Report an error
The week in bills
One email, Sunday: what Congress passed, what it changed, and who paid.
One link arrives to confirm the address. It is used for this email and nothing else, and every issue carries an unsubscribe link.