Explore the influence map → · Follow Sanders in the full network →
A real policy channel
Bernie Sanders and Greg Casar connect this investigation to a concrete proposal for federal control of advanced AI development. ControlAI publicly says it consulted on their framework. That is a direct working connection to an organization already prominent in this corpus. Its exact drafting contribution, contracting legal entity and any fee remain unknown. The September 3 sponsor announcement calls the legislation forthcoming; this inquiry did not recover a filed bill number or complete congressional statutory text.
Sanders and Casar announce forthcoming Ban Artificial Superintelligence Act · First Bill Introduced to Ban Superintelligent AI
What authority would be created
The public summary proposes a permanent ASI prohibition, a temporary advanced-AI pause, a cabinet-level regulator and an expert advisory board. It specifies monitoring, supervised capability removal and destruction, corporate dissolution and up to 20 years imprisonment. It also pursues international prohibition through agreements, allied coordination and export controls. These are proposed powers, not existing law.
Ban Artificial Superintelligence Act — sponsor release summary
The definition is part of the power
The sponsor summary covers systems that match or exceed human cognitive performance across a broad range of tasks, or can plan and execute human disempowerment, including undermining the US government. Easily modified systems are included. Who interprets those categories, what proof is required and how decisions can be challenged are therefore central questions. The one-page summary cannot settle the full procedure.
Ban Artificial Superintelligence Act — sponsor release summary
Consultation does not mean identical policy
Miotti explicitly describes Sanders’s temporary advanced-AI pause as broader than ControlAI’s preferred approach. ControlAI has a separate discussion draft. In its inspected definition, three characteristics must occur together; the Sanders summary instead uses alternatives. The ControlAI draft also specifies court procedures for continued seizure and destruction. It leaves congressional identifiers and the responsible Secretary blank. Its mutable PDF must not be substituted for Sanders’s missing full text, or assumed unchanged since the landing page’s June 24 date.
First Bill Introduced to Ban Superintelligent AI · ControlAI US ASI Bill / Artificial Superintelligence Prohibition Act of 2026 · Discussion draft: a bill to ban superintelligent AI and regulate its precursors
Expert access has a history
Sanders said in February that Geoffrey Hinton had recently spoken to him about AI becoming smarter than humans. An April 29 panel provides an earlier Tegmark connection. Sanders’s official video page describes bringing US and Chinese scientists together; contemporary reporting names Tegmark, David Krueger, Zeng Yi and Xue Lan. FLI’s roster separately identifies Tegmark as Founder and Chair. Individual public participation is not proof of FLI drafting or financing.
Sanders to Meet with AI Leaders in California · The Existential Threat of AI and the Need for International Cooperation · Sanders Convenes Chinese and American Computer Scientists to Warn of Threat from AI · Our people
A planned briefing is not completed attendance
Axios named Hinton, Tegmark and Cotra for a planned September 16 Senate briefing. Sanders told CNN he invited every senator, but the rush transcript twice refers to Thursday. The invitation letter was not obtained, leaving the exact date unresolved. These records do not appoint anyone to the proposed advisory board.
Sanders convenes senators for AI briefing amid mounting alarm · The Source with Kaitlan Collins, September 9, 2026
How research enters the political argument
Sanders’s September 3 essay cites Ajeya Cotra and Dwarkesh Patel to make the case for a pause. METR identifies Cotra as staff and a coauthor of its August incident investigation; it says OpenAI did not pay for that assessment. The essay footer credits Friends of Bernie Sanders. These are observable steps in communication: research, public interpretation, citation by a politician and distribution. They do not establish that the cited people endorse every proposed penalty, or independently verify the incident described.
Pause AI Development NOW · Brief independent investigation of agents' behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
The campaign mechanism is visible
Civic Shout hosts a petition advocating this exact Act. Its joint-action preview names College Democrats of America and Left Click and offers recruiting participants $2.25 in advertising credits per recruited signature. It promises delivery to Senate Commerce and House Science committee leadership, then letters to members. This identifies an incentive and pressure channel. It does not establish delivered signatures, credit redemption, cash payments or instructions from the sponsors’ offices.
Stop Big Tech from Building Uncontrollable AI — joint action preview
A poll measures its question
Data for Progress’s September 4–9 weighted web panel of 1,332 likely voters found 68% support and 25% opposition for the described combination of a pause and permanent prohibition. The questionnaire neither names the sponsors nor presents prison terms, corporate dissolution, export controls or the new agency’s architecture. It measures support for selected policy components rather than informed approval of the complete legislation. No commissioner or funder was identified in the inspected materials.
Polling on AI Development Regulation (September 2026) · Data for Progress — AI development pause questionnaire and crosstabs
Which friends support which measure
Casar chairs the Congressional Progressive Caucus; Sanders is listed as its Senate member. That is a political relationship, not caucus-wide adoption. Casar’s August tax-and-jobs proposal has named support from AFT, AFSCME, Groundwork Action and Demand Progress Action. Sanders and Alexandria Ocasio-Cortez jointly announced the March data-center moratorium; National Nurses United supports that measure. Those earlier endorsements stay attached to their actual proposals.
Caucus Members · Casar Leads Introduction of New Bill To Protect Workers From Threat of AI Mass Unemployment · Sanders, Ocasio-Cortez Announce AI Data Center Moratorium Act · National Legislation — Artificial Intelligence Data Center Moratorium Act
The money search has a defined boundary
Friends of Bernie Sanders, C00411330, is the principal Senate committee. FEC reports $24.93 million receipts, $23.78 million contributions and $22.84 million ending cash for January 2025 through June 2026. Of individual contributions, $17.27 million is unitemized and $6.51 million itemized. The amended July filing corroborates the cash balance. Its election-cycle totals use a different basis and should not be mixed with the two-year overview.
Friends of Bernie Sanders — FEC 2025–2026 committee overview · Friends of Bernie Sanders — July quarterly amendment FEC-2003883
No donation bridge established in the checked records
Separate queries for Moskovitz, Tegmark, Tallinn and Tuna returned no disclosed entries for that committee and transaction period. Selected AI-safety employer queries also returned none. This is not a lifetime donor audit, a review of Casar’s finances, or proof that no outside resources supported advocacy. Unitemized contributions, spelling variants, other committees and earlier periods limit the result; available committee reports also predate September. No financial edge from the named AI-safety donors to Sanders has been added.
FEC Schedule A — C00411330 / MOSKOVITZ / 2025–2026 · FEC Schedule A — C00411330 / TEGMARK / 2025–2026 · FEC Schedule A — C00411330 / TALLINN / 2025–2026 · FEC Schedule A — C00411330 / TUNA / 2025–2026
Conflict remains visible
ITIF publicly opposed the Sanders–Casar prohibition, emphasizing innovation and competition. Gary Marcus also opposed its breadth while arguing for regulation; he favored an alternative formulation by Anthony Aguirre. Casar’s office meanwhile reports actual correspondence demanding security records from OpenAI and Anthropic. The record contains disagreement among regulators’ advocates and conflict between lawmakers and companies. Treating all of them as one camp would erase useful evidence.
Sanders-Casar AI Ban Would Sacrifice American Innovation and Competitiveness, Says ITIF · The new Sanders-Casar Ban Artificial Superintelligence Act – and why I oppose it · Casar Responds to OpenAI, Anthropic, Demands Greater Transparency About Major Security Lapses
What this changes in the larger picture
Our interpretation: the network trades different resources. Advocacy organizations offer policy work, researchers offer technical authority, campaign platforms recruit attention and signatures, and legislators can turn proposals into compulsory rules. None of those currencies needs to be a campaign donation to matter. But contact alone does not prove control, and the public record contains real disagreement over the desired rules. This branch extends the lens from access to proposed state authority.
First Bill Introduced to Ban Superintelligent AI · Ban Artificial Superintelligence Act — sponsor release summary · Stop Big Tech from Building Uncontrollable AI — joint action preview
Evidence worth finding next
The highest-value next artifacts are the filed bill and revision history; sponsor-office correspondence identifying ControlAI’s contribution; expert invitation and attendance records; proposed appointment, review and appeal provisions; any bill-specific consulting or grant terms; and campaign delivery or credit-redemption records. These could distinguish policy influence, paid implementation and actual exercise of authority. None is assumed to exist in a particular form or to contain wrongdoing.