S.Amdt. 6188

This amendment, submitted by Mr. Warner, aims to stop the Federal Government from buying or using specific artificial intelligence models that fail to prevent the creation of harmful content.

Key Changes and Additions:

• Prohibition on Use: Federal agencies and their contractors are banned from procuring or using "covered applications." These are AI models that do not meet NIST standards regarding non-consensual intimate imagery and child sexual abuse material, or those that Federal courts have determined generate child pornography or non-consensual intimate images of adults.

• Safety Benchmarks: The Director of the National Institute of Standards and Technology (NIST) must create and update performance benchmarks to stop the generation of obscene, degrading, and abusive content.

• Enforcement and Review: Agency heads must remove covered applications from their systems. The General Services Administration (GSA) and agency procurement officers must regularly review products to ensure no covered applications are being used.

• Developer Rules and Liability:

  • Developers are given a limited "safe harbor" from certain criminal laws if they possess child pornography solely to create technical controls to prevent it, provided they follow strict reporting and deletion guidelines.
  • It is prohibited to make or sell technology specifically designed to bypass these technical controls.
  • The amendment creates a private right of action, allowing injured individuals or developers to sue in Federal court for damages (including statutory damages of $50,000) and injunctions.