Gottheimer Announced New Bipartisan AI Safety Bills
The proposed legislation targets federal procurement and mandatory security reviews for advanced AI systems.
Updated on Sept. 18, 2026 in Artificial Intelligence
Live Poll
Should the government require mandatory safety reviews for all new advanced artificial intelligence models?
Representative Josh Gottheimer announced two bipartisan bills aimed at increasing oversight of artificial intelligence. The American AI Security Act requires government review of advanced AI models, while the China FIREWALL Act restricts the use of Chinese-developed AI on federal devices.
Why it matters
This legislative push addresses rising national security concerns regarding the rapid deployment and provenance of artificial intelligence models. It signals a move toward formalizing federal vetting processes for software that could influence government operations.
The American AI Security Act establishes a 30-day government review process for advanced AI models before public release, which may be extended by an additional 30 days. The National Security Agency is designated to lead this mandatory evaluation phase.
The players
Josh Gottheimer
Representative for New Jersey's 5th congressional district who focuses on national security and technology policy.
National Security Agency
The United States intelligence agency tasked with global monitoring, collection, and processing of information for intelligence and counterintelligence purposes.
The details
The American AI Security Act creates a mandatory framework requiring developers to submit advanced models for a security assessment by the National Security Agency. Separately, the China FIREWALL Act targets procurement by prohibiting federal agencies from purchasing software reliant on Chinese-developed AI models. This bill also mandates the removal of Chinese-developed AI models with open weights—publicly accessible software code for AI architectures—from government-issued hardware.
Timeline
September 18, 2026: Representative Gottheimer announced the proposed AI safety legislation.
The Tech Race
This legislation extends the safety standards established by the 2023 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence by moving oversight from executive guidance to mandatory legislative review. It marks a shift toward stricter government-led vetting of AI supply chains and model architectures compared to current voluntary industry frameworks.
The proposed bans on specific AI models and software procurement will primarily impact federal employees and government contractors who manage IT procurement. If passed, these rules will force agencies to re-evaluate their software stack to ensure compliance with new security provenance standards.
The takeaway
The move signals a hardening stance on the integration of foreign-developed models into critical infrastructure. Watch for the upcoming legislative debate in Congress to see if these bills move toward a committee markup or floor vote.
Further reading
For broader context on current regulatory efforts, visit Artificial Intelligence.
Live Poll
Should the government require mandatory safety reviews for all new advanced artificial intelligence models?
