When will AI biosecurity safeguards become U.S. law? — AI biosecurity controls

A qualifying federal law must require evaluations or safeguards for AI models or systems that specifically address biological misuse or harm. Mandatory biosecurity testing or evaluations qualify, as do controls on access, capabilities, security, deployment, or procurement tied to biological risk. The requirement may be part of a broader AI law or apply only to a defined sector or federal procurement. Studies, voluntary standards, funding, and gene-synthesis, laboratory, or biological-data rules without an operative AI-model or system requirement do not qualify. General AI or cybersecurity requirements and incident reporting alone are insufficient without the defined biosecurity evaluation or safeguard. A later effective date does not delay qualification. An example of this is Section 1654 of S. 4784, the Senate FY2027 defense authorization bill.