Anthropic privately circulates draft federal frontier AI bill to lawmakers, Politico reports
For the general counsel at a regional bank
Anthropic has been privately circulating a draft bill to lawmakers on Capitol Hill that would set federal rules for frontier AI systems, according to Politico reporting as relayed by Crowdfund Insider on Oct. 18; the draft itself is not among the sources.1
Politico obtained the confidential proposal and spoke with three people familiar with the discussions, and said it closely tracks Anthropic's internal Advanced AI Framework, which the company has promoted to federal and state officials since mid-2025.1 An Anthropic spokesperson described the text as reflecting "sensible policies" the firm believes Congress should enact.1
Authorities could act to block a model's release or continued use only for a foreseeable and imminent risk of more than $1 billion in damages or the death or serious injury of more than 50 people, and only if the harm came through a biological weapon, a model escaping its developer's controls, or a cyberattack.1
Independent evaluators registered with the Department of Commerce would assess frontier labs and their most advanced models periodically, and developers could pay them as long as payment did not depend on the outcome of a review.1
If evaluators found an "imminent enumerated risk," they would have to notify Commerce within 72 hours, and the department could then petition a federal judge in the District of Columbia to suspend the model; the draft would bar any other federal department, agency or official from suspending development, deployment or internal use.1
The draft would preempt some state requirements on risk disclosure, third-party evaluation and critical-incident reporting, and would require safety frameworks, transparency reports, whistleblower protections and incident reporting to Commerce.1
If enacted as described, banks and fintechs that depend on a frontier model would face a narrow, court-based path to any federal suspension, and harms below the thresholds or outside the three pathways would trigger no federal shutdown. Preemption could also change which state disclosure and incident-reporting rules apply to AI vendors, so vendor contracts and continuity plans may need to account for both.
Our take. Under this draft, the only federal route to stopping a model your bank runs on is Commerce asking a DC federal judge, over a $1 billion risk. Does your bank examiner count as any other federal agency?
Action. Legal: Map frontier-model vendor contracts for suspension and termination rights, and list which state incident-reporting duties the draft would preempt.
What our records say
Claude's posture toward banks is direct sales of seats and API access, plus deeper engagements added in 2026.1 Anthropic sells banks Claude seats and API access directly and through cloud marketplaces, and in 2026 added embedded engineering with Goldman Sachs and FIS, restricted cyber-defence access to its Mythos model for a few large banks, and a services firm co-founded with Goldman Sachs.1
From our records
- Claude (Anthropic)vendor · banks using it (8) Goldman, JPMorgan, BankUnited · alternatives Microsoft Copilot, ChatGPT Enterprise, Gemini Enterprise
- Commerce Bancsharesbank · latest Nolan & Associates acquisition completed (Oct 2026)
- Columbia Banking Systembank · latest Barclays conference: AI answers most client messages (Sep 2026) · vendors Glia · enforcement 1 action since 2016; latest FDIC order to pay civil money penalty, $1.8 million (May 2021)