Sens. Peter Welch, D-Vt., and Ben Ray Lujan, D-N.M., have introduced the Trustworthy by Design Artificial Intelligence Act, which seeks to establish best practice guidelines for federal agencies’s efforts to develop AI systems.
Under the Senate bill, the National Institute of Security and Technology would be required to formulate a framework for the proactive and systematic management of AI risks, Welch said in a Dec. 23 press release. NIST would also be directed to create definitions of trustworthiness evaluation criteria and identify system components that must be evaluated when developing AI capabilities.
Table of Contents
Ethical Deployment of AI Tech
Welch explained that his proposal would ensure that the United States will maintain its leadership in the ethical deployment of AI technologies and the creation of guidelines to provide users with trustworthy AI tools.
“With great power comes great responsibility, we need to ensure AI is used safely,” the lawmaker stressed.
If passed into law, the TBD AI Act would guarantee that AI models in the country are reliable, safe and secure. The measure would also reduce the potential biases affecting legally protected classes and risks associated with system development. In addition, the legislation would mandate NIST to provide periodic public updates on the program’s implementation.
Transparent AI Systems Evaluation
Several stakeholders and organizations are supporting the act, including the Transparency Coalition. The group welcomed the legislative initiative, noting it would promote transparency in evaluating AI systems and set the standard for trustworthy AI implementation and deployment in federal agencies.