At a US House hearing, Treasury Secretary Scott Bessent said AI labs should get no liability exemptions and called for more open-source models built in the US
Matt Bracken /FedScoop:NEW
Context & Ripple Effects
The administration had already cast AI model development as a strategic competition, including Treasury’s review of whether Chinese models were distilled from US systems. Bessent’s hearing position joins that competitive framing to a domestic argument for US-built open-source models.
The policy backdrop is becoming more concrete: Senate negotiators are weighing a duty-of-care proposal that could block unsafe model releases, while OpenAI had rejected an Illinois liability shield and backed stronger safety legislation. Bessent’s stance puts a Treasury voice against special liability treatment even as federal rules remain unsettled.
First-order effects
- AI labs lose a potential Treasury advocate for liability exemptions, strengthening the public expectation that they bear responsibility for harms tied to their models.
- US open-source developers gain high-level political backing as Bessent frames domestically built models as part of the response to Chinese competition.
Second-order effects
- Senate advocates of a duty of care can point to Bessent’s rejection of exemptions when arguing that safety obligations should attach to AI companies rather than be traded away for industry support.
- Labs seeking favorable federal treatment face pressure to separate their competitiveness case from requests for legal protections, while open-model proponents gain a clearer industrial-policy rationale.
Third-order effects
- AI policy is taking on a two-track structure: governments may promote domestic model availability for strategic reasons while resisting blanket limits on developer accountability.
- If that structure holds, the advantage will shift toward labs able to present both a credible safety posture and a role in US model competition, rather than treating those goals as substitutes.
The trend: US AI policy is increasingly pairing strategic support for domestic model ecosystems with demands that developers retain responsibility for safety and downstream harms.