Anthropic pitches draft federal frontier AI safety bill to US lawmakers, Politico reports
Anthropic has been circulating a draft bill among US lawmakers that would let the government block dangerous frontier AI models, subject to a judge's approval, according to Politico. Anthropic confirmed the effort and says the bill reflects its published policy framework.
1 / 3
The story, neutrally told
Centre · 1Anthropic representatives have quietly pitched Capitol Hill lawmakers on a draft bill to regulate frontier AI safety and security, Politico reported on 16 October, citing three people with knowledge of the efforts. CNAN “Anthropic representatives have quietly pitched Capitol Hill lawmakers on a draft bill to regulate frontier AI safety and security, Politico reported on Friday, citing three people with knowledge of the efforts.” Read at CNA ↗ Right · 1Anthropic confirmed it is circulating the draft among US lawmakers, to address the lack of federal legislation on the risks of the technology, according to AFP. The Straits TimesRC “Anthropic, the creator of the AI assistant Claude, confirmed on Oct 16 it is circulating a draft Bill among US lawmakers” Read at The Straits Times ↗ Mixed · 2Under the draft, the government could block a model's release or use only if it poses a foreseeable, imminent risk of more than US$1 billion in damages, or of killing or seriously injuring more than 50 people. The risk must stem from a biological weapon, the model evading its developer's controls, or a cyberattack. CNAN “The risk must stem from the creation or release of a biological weapon, the model evading its developer’s controls or a cyberattack, Politico reported.” Read at CNA ↗ The Straits TimesRC “Washington could block a model’s release or use if there is an imminent risk of killing or seriously injuring more than 50 people” Read at The Straits Times ↗
Centre · 1Independent evaluators, registered with the Commerce Department, would periodically evaluate each lab and its frontier models. Labs could pay them, provided the payment is not tied to the outcome of the evaluation. CNAN “The labs could pay these evaluators, provided the payment is not tied to the evaluation's outcome, the report said” Read at CNA ↗ Centre · 1If evaluators find an "imminent enumerated risk", they must report it to Commerce within 72 hours of completing their assessment. Commerce could ask a judge to halt a model, and a federal judge for the District Court for the District of Columbia could order the suspension. The language reportedly bars any other department, agency or officer from suspending development, deployment or internal use. CNAN “they must report it to Commerce within 72 hours of completing their assessment, Politico reported”“the language explicitly prohibits any other US government department, agency, or officer from suspending development, deployment, or internal use” Read at CNA ↗ Mixed · 2An Anthropic spokesperson said the proposal "grants the federal government the ability to block or deter dangerous deployments of the most powerful models". The spokesperson said it also requires frontier developers to test the most powerful models, be transparent with the public about their findings, submit to independent evaluation, and maintain a robust security programme. Reuters, via CNA, quoted the spokesperson as saying the bill reflects proposals in Anthropic's "Advanced AI Framework" released earlier this year. The Straits TimesRC ““It requires frontier developers to test the most powerful models, be transparent to the public about their findings, submit them to independent evaluation, and maintain a robust security programme,” the spokesperson said.” Read at The Straits Times ↗ CNAN “An Anthropic spokesperson told Reuters the Bill represents proposals outlined in Anthropic's "Advanced AI Framework" released earlier this year.” Read at CNA ↗
Centre · 1The Department of Commerce did not immediately respond to a Reuters request for comment. CNAN “The Department of Commerce did not immediately respond to a Reuters request for comment.” Read at CNA ↗ Mixed · 2The proposed legislation would override certain state laws, such as those in force in California. Anthropic wrote in June that "absent a strong federal law, states should retain full authority to legislate". CNAN “The proposed legislation would also override certain state laws, such as those in force in California.” Read at CNA ↗ The Straits TimesRC “Anthropic wrote in June that “absent a strong federal law, states should retain full authority to legislate”.” Read at The Straits Times ↗ Right · 1Charlie Bullock of the Institute for Law & AI think tank called the proposal "pretty reasonable" and "more than other frontier labs have done", but said on X that he considers it "weaker" than the FRONTIER Act, a competing bill introduced in the House of Representatives this summer. The Straits TimesRC “But he said on X he considers it “weaker” than the FRONTIER Act, a competing bill introduced in the House of Representatives this summer.” Read at The Straits Times ↗
Right · 1The wire reports give the political context: US AI policy under President Donald Trump has relied mainly on voluntary industry pledges and executive orders, and Congress, which has yet to pass a federal AI law, is out of session until after next month's midterm elections. The Straits TimesRC “Congress, politically divided and heavily lobbied by Big Tech, has yet to pass a federal AI law, and is out of session until after next month’s midterm elections.” Read at The Straits Times ↗ Right · 1The reports also note that Anthropic was banned from the Pentagon by the Trump administration in February and was forced by Washington in June to suspend its most advanced models for nearly three weeks. They add that CEO Dario Amodei called in September on the industry to slow down, fearing AI agents could take control of "the entire internet" within six to 12 months, and that Trump has dismissed calls to slow the technology as "negative forces". The Straits TimesRC “Anthropic, which was banned from the Pentagon by the Trump administration in February, was forced by Washington in June to suspend its most advanced models for nearly three weeks.”“Trump has dismissed the numerous calls to slow down the technology as “negative forces”.” Read at The Straits Times ↗
Every sentence links to the reporting it rests on. The pill in front of each says where its sources sit: Left, Centre or Right when one side supplies at least half of them, Mixed when they are evenly split. The number is how many outlets it cites.
Left1 outlet
No left outlet in our sources has covered this story yet.
Centre1 outlet
- Framing
- Reuters-based CNA reports lead with the Politico scoop and set out the bill's mechanics in detail: risk thresholds, evaluators, the 72-hour reporting rule and the judicial route.
- Emphasis
- Bill mechanics (evaluator registration, payment rules, Commerce's role, D.C. District Court), plus the Anthropic spokesperson and Commerce non-response.
- Leaves out or plays down
- The shorter version leaves out the political context, state-law preemption and expert reaction.
- Charged language
- “quietly pitches”
Right1 outlet
- Framing
- The Straits Times, carrying AFP copy, frames the story as Anthropic confirming it is circulating a draft to fill the gap in federal AI law, set against Trump-era policy and recent AI-agent security incidents.
- Emphasis
- Anthropic's confirmation, the judge-approval requirement, Trump's stance, and Anthropic's run-ins with Washington.
- Leaves out or plays down
- Omits the evaluator and 72-hour reporting details and the Commerce non-response.
- Charged language
- “negative forces”
- For example
-
“Anthropic’s proposed law grants the federal government the ability to block or deter dangerous deployments of the most powerful models.” — The Straits Times
“Security became a hot topic after recent revelations of security incidents caused by AI agents.” — The Straits Times
What every side reports
- Politico reported on 16 October that Anthropic is pitching a draft frontier AI safety bill to US lawmakers.
- The draft would allow blocking a model over imminent risk of more than US$1 billion in damages or of killing or seriously injuring more than 50 people.
- An Anthropic spokesperson said the bill lets the government block or deter dangerous deployments of the most powerful models.
- Congress has not passed a federal AI law and is out of session until after the midterm elections.
- The bill would override some state laws, including California's.
Where accounts differ
-
Whether Anthropic's account matches the leaked draft details
- Centre
- CNA, relaying Reuters, says the bill 'represents proposals outlined' in the Advanced AI Framework and notes Commerce had not responded.
- Right
- The Straits Times (AFP) stresses that a block requires a judge's approval and that Anthropic confirmed circulating the draft; it does not mention the framework.
Anthropic organisation
Anthropic confirmed circulating the draft and says it lets the government block or deter dangerous deployments while requiring testing, transparency, independent evaluation and security programmes. It ties the bill to its Advanced AI Framework.
“A company spokesperson told AFP that the proposal “grants the federal government the ability to block or deter dangerous deployments of the most powerful models”.” — The Straits Times
“An Anthropic spokesperson told Reuters the Bill represents proposals outlined in Anthropic's "Advanced AI Framework" released earlier this year.” — CNA
AI regulation topic
The bill enters a debate where federal AI policy relies on voluntary pledges and executive orders, and a competing House bill, the FRONTIER Act, is seen by one analyst as stronger.
“AI policy in the United States under President Donald Trump has relied primarily on voluntary industry pledges and executive orders” — The Straits Times
Left1 article
Centre2 articles
-
CNAN ·
Anthropic quietly pitches its own AI safety Bill to US lawmakers
Neutral Fuller Reuters-based report detailing the bill's terms, Anthropic's response and political context.
-
CNAN ·
Anthropic quietly pitches its own AI safety bill to lawmakers, Politico reports
Neutral Short Reuters bullet-point summary of the Politico report's bill provisions.
Right1 article
-
The Straits TimesRC · · via AFP
Anthropic proposes law regulating AI to US lawmakers
Neutral AFP copy centred on Anthropic's confirmation and the judge-approval limit, with Trump-era policy context.
- 16 Oct 23:37 First CNAN Anthropic quietly pitches its own AI safety Bill to US lawmakers
- 16 Oct 23:37 +1m CNAN Anthropic quietly pitches its own AI safety bill to lawmakers, Politico reports
- 17 Oct 08:29 +8h 52m The Straits TimesRC Anthropic proposes law regulating AI to US lawmakers via AFP
- 17 Oct 08:42 +9h 5m The HinduLC Anthropic proposes law regulating AI to US lawmakers via AFP
Times are when each article was published, or when we first saw it if the outlet gave no time.
Ask about this story
Loading…