AOC、カンナ氏、その他8人の下院民主党議員が人工超知能を禁止する法案を支持

AOCやロー・カンナを含む下院民主党10議員は、人工超知能を禁止し、連邦AI省を創設するサンダース・カザール法案を支持している。

Ten House Democrats, including Reps. Alexandria Ocasio-Cortez of New York and Ro Khanna of California, have publicly backed legislation that would ban artificial superintelligence, giving the Sanders-Casar proposal its first visible bloc of House support three weeks after it was introduced, the Washington Examiner reported Wednesday night.

The bill, introduced in the Senate by Sen. Bernie Sanders (I-VT) and in the House by Rep. Greg Casar (D-TX), would ban artificial superintelligence — defined in the legislation as an AI model that "exceeds human cognitive performance and capabilities across most domains or tasks" — and create a federal Department of Artificial Intelligence to regulate advanced systems.

When AI Buzz Wire covered the bill's introduction in early September, its fate looked like a messaging exercise: sweeping in ambition, short on co-sponsors, and facing near-certain opposition from both industry and the majority party. The new list of House backers does not change the arithmetic of passage, but it changes the politics. A ban on superintelligence is no longer a two-person cause; it is now the official position of at least a dozen members of Congress.

Who is backing the ban

According to the Washington Examiner, several "prominent House leftists" joined Ocasio-Cortez and Khanna in supporting the measure, which the outlet describes as a ban on artificial superintelligence paired with the creation of a new federal department. The full list of ten House Democrats was not itemized in the report's public text, but the presence of Ocasio-Cortez and Khanna — two of the most recognizable names in the House progressive caucus — signals that the proposal has moved from a Sanders-Casar statement project into a genuine factional position within the Democratic Party.

It also aligns with the left flank's broader AI agenda. Sanders and Ocasio-Cortez have previously joined forces on proposals for a moratorium on AI data center construction, and Khanna has positioned himself as one of the chamber's most active legislators on tech accountability. The superintelligence ban extends that framework from where AI is built to whether its most advanced form should be built at all.

What the bill would do

The legislation's central provision is a permanent legal line: no entity would be permitted to build or deploy a system that exceeds human cognitive performance across most domains. The definition matters — by targeting capability thresholds rather than specific techniques, the bill aims to regulate a class of system rather than a snapshot of today's technology.

The proposed Department of Artificial Intelligence would house federal oversight of advanced models, a structural upgrade from the current patchwork of voluntary commitments, safety institutes, and sector-by-sector regulation. Critics have called the department an enormous expansion of federal power over software development; supporters argue that only a dedicated agency can meaningfully evaluate systems whose failure modes are national in scale.

As reported when the bill was introduced, the proposal also calls for a temporary pause on advanced AI development — a demand Sanders framed as giving society time to decide, deliberately, whether to cross the superintelligence threshold. International coordination is part of the pitch: Sanders has urged governments worldwide to work together to prevent such systems from being built, framing runaway AI as a global coordination problem.

The political math

The bill faces steep odds. Republicans control the Senate, the White House has pushed a light-touch, infrastructure-first AI agenda, and industry groups have warned that banning capability milestones would cede ground to China — an argument that gained force this year as Chinese labs shipped competitive open-weight frontier models. Even among Democrats, the proposal sits well outside the party's mainstream, which has coalesced around transparency rules, safety testing, and market structure rather than hard capability limits.

しかし、この法案は実際には今議会を対象としたものではない。それは次の議題、そして私たちの AI 政策ニュースの報道が年間を通じて追跡してきたように、業界自体の軌跡を目指しています。それぞれの不正エージェント事件、それぞれの安全上の欠陥が明らかになり、政府によるフロンティアモデルの見直しは、禁止の支持者たちに新たな証拠を与えている。この夏だけでも、議員らは不正AIエージェントに関する証言をOpenAIとAnthropicに要求し、州議会は独自のフロンティアモデル審査体制を進め、ホワイトハウスはどの国のテスターが最初に新しいモデルにアクセスできるかについて自ら介入し始めた。

これが、下院民主党議員10名が超情報機関の禁止がその名に値すると判断した背景である。 2026年までの可決がどんなに遠ざかろうとも、この法案は現在、指標として機能している。法律用語で言えば、機械の認知が超えてはならない一線があり、新たな連邦部門がそれを施行すべきであると両院ともに初めて正式に制定された法案である。

この指標がより多くの共同スポンサーを集めるのか、それとも中間選挙後に消えるのかは、不正エージェントの夏が実際にAIに対する厳しい制限を求める議会の志向をどれほど変えたかについて多くを語ることになるだろう。