Sen. Peter Welch, D-Vt., has introduced a bill that would require the Department of Defense to report to Congress on risk assessments of the use of artificial intelligence in defense systems as part of a push to improve DOD’s ethical technology leadership and transparency.

Welch’s office said Tuesday the proposed Artificial Intelligence Weapons Accountability and Risk Evaluation Act, also known as the AWARE Act, would establish a risk assessment process that evaluates AI technologies against several factors, including bias, bias towards escalation, cybersecurity, privacy, dependability and risk of civilian harm.

The measure would apply the risk assessment to any AI-enabled weapons, targeting or decision support tools, catalog the covered technologies and assessment outcomes in a risk ledger that is unclassified to the extent possible but may include a classified appendix and annotate which of the tech platforms have been shared, exported or used by a foreign individual or government.

Under the bill, DOD should perform an annual risk assessment of all covered AI-enabled systems and conduct a new risk assessment if a weapons review process is initiated, when a technology is procured, or when there is an update to an underlying AI model.

The legislation includes a requirement to report to Congress on an annual basis on the progress of implementing the measure, the ledger and a report on the findings.

“We know that AI-enabled weapons are exacerbating harm to civilians in warzones, and without effective oversight these systems could dangerously reshape our future. The United States has an obligation as a leader in AI innovation to also lead in AI ethics,” said Welch.

“We must be ready to answer questions about accuracy, security, privacy, and harm when we use AI-enabled weapons—and we need better data to do that,” added the Senate Commerce Committee member.

Story Written by Jane Edwards, ExecutiveGov

Story Link: https://executivegov.com/2024/10/senate-bill-risks-ai-defense-systems/