Bill · S. 5061· Senate(119-S-5061)
Secure A.I. Development Act of 2026
- Introduced
- 7/21/2026
- Last activity
- 7/21/2026 — Read twice and referred to the Committee on Commerce, Science, and Transportation.
- Sponsored by
- Mark Warner
- Cosponsors
- None
- Introduced
- In Committee
What this bill would do
This bill, known as the "Secure Artificial Intelligence Development Act of 2026," aims to improve the tracking and processing of security and safety risks related to artificial intelligence. It establishes an Artificial Intelligence Risk Board to develop technical evaluations and best practices for high-performance AI models. The legislation requires providers of "frontier" AI models to grant the National Security Agency access to those models before their public release. Additionally, it creates a registry for these models and establishes a voluntary database for tracking safety and security incidents. The Act also directs updates to national vulnerability databases and establishes a pilot program for sharing intelligence regarding supply chain risks.
Yes· 3 of 6
Does this better protect the country from computer attacks?+8
The bill strengthens national cybersecurity by establishing a risk board, updating vulnerability databases, and facilitating intelligence sharing regarding AI threats.
Does this make what the government does easier for the public to see?+3
The act creates a publicly accessible database for tracking AI safety and security incidents.
Does this protect people from being watched or searched?+2
The bill includes provisions to ensure anonymity and protect privacy when reporting AI incidents.
The opposite
Would business owners be better off under this?-5
AI developers face heavy regulatory burdens, mandatory model registry requirements, and high daily fines for non-compliance.
Does this make it easier to run a small business?-3
The bill imposes significant compliance, registration, and potential financial penalties on AI providers.
Does this make trade with other countries easier?-2
Mandatory registration and access requirements for frontier models may create friction in international AI commerce.
Who it would affect, and how
Potential Positive Impacts for People
-
Enhanced Safety Oversight: The creation of the Artificial Intelligence Risk Board will allow for the development of technical evaluations to determine which AI capabilities pose serious risks to national security, economic security, or public health and safety (Section 3(b)(6)(A)).
-
Standardized Safety Information: The Board will develop best practices, including standard formats for publishing model cards with technical details, which can help provide transparency regarding AI systems (Section 3(b)(6)(A)(ii)(I)).
-
Improved Risk Tracking: The establishment of a publicly accessible database for artificial intelligence security and safety incidents will allow for the tracking of events that could endanger human life, health, property, or the environment (Section 2(3), Section 4(a)).
-
Cybersecurity Strengthening: The bill directs the National Institute of Standards and Technology and the Cybersecurity and Infrastructure Security Agency to update vulnerability management processes and the National Vulnerability Database to better incorporate artificial intelligence security vulnerabilities (Section 5(b)).
-
Supply Chain Security: A pilot program will be established to facilitate the secure sharing of intelligence and threat information to help protect the supply chains used for training and maintaining artificial intelligence models (Section 7(c)).
Potential Negative Impacts for People
-
Financial Penalties for Providers: Providers of frontier artificial intelligence models that fail to provide access to the National Security Agency prior to public release can be fined at a rate of not less than $100,000 per day (Section 3(e)(3)).
-
Strict Compliance Requirements: To avoid enforcement actions, providers must demonstrate compliance by withdrawing a model from commerce and providing the National Security Agency with access to the model, including its weights and configuration files (Section 3(e)(4)).
-
Mandatory Registration Burdens: Providers of frontier artificial intelligence models are obligated to register their models with a specific registry before they can be introduced into interstate or foreign commerce (Section 3(d)(3)).
Groups and Interests Ranking
Every group the model scored this bill for, from -10 to +10. Hover a score for the reason it gave.
- National Security And Foreign Policy
- Improves Cybersecurity Defense+8Improves International Trade-2
- Demographic And Interest Groups
- Good For Business Owners-5
- Governance And Ideology
- Improves Government Transparency+3Improves Privacy Rights+2
- Economy And Finance
- Improves Small Business Support-3
- Education And Technology
- Increases Scientific Research Funding+2
- Environment Energy And Infrastructure
- Improves Public Land Management+1
- Social Issues And Justice
- Improves Poverty Reduction+1
- 7/21/2026
- Read twice and referred to the Committee on Commerce, Science, and Transportation.
- Introduced in Senate