Senate AI Group Said to Be Focused Despite Briefing Attendance Drop-Off

There has been a “little drop-off” from the first to the third briefing, but Wednesday still had a “good showing,” Sen. Todd Young, R-Ind., told us Thursday. About 70 senators attended the first, and about 60 attended the second. Between 75 and 80 staffers and senators attended Wednesday, said Sen. Mike Rounds, R-S.D. Schumer formed an AI working group with Young, Rounds and Sens. Martin Heinrich, D-N.M. Turnout Wednesday was “not as good” as it was for the second briefing, but it remains an “educational” process, said Rounds: “We don’t want to do harm. We want to do whatever we can to expand the opportunity for this tool to be developed in the U.S. and stay ahead of our adversaries.”

Senate Intelligence Committee ranking member Marco Rubio, R-Fla., said his panel has been hosting educational sessions on the topic for more than a year. Committee members previously heard from two of the three presenters at last week’s briefing, he said. Attendees heard perspectives from federal agencies, what research has been done, what research is needed and what resources and authorities agencies need to ensure the U.S. continues to lead the way, said Young.

Senate Intelligence Committee Chairman Mark Warner, D-Va., said he attended the first 20 minutes of Wednesday’s briefing and Schumer is “doing the right thing.” He said getting educated on AI is often “two steps forward, one step back.” Warner estimates he spends about 10-15 hours a week on the topic. When you spend that much time on something, often “you feel like you’re on an upward path of learning,” but AI is increasingly complex, he said.

Members of the Congressional AI Caucus introduced legislation Friday that attempts to give AI researchers better access to data. Reps. Anna Eshoo, D-Calif.; Michael McCaul, R-Texas;. Don Beyer, D-Va.; and Jay Obernolte, R-Calif., introduced the Creating Resources for Every American to Experiment with Artificial Intelligence Act (Create AI Act). The bill would establish the National Artificial Intelligence Research Resource as a central hub for research access. Young, Rounds and Heinrich introduced companion legislation with Sen. Cory Booker, D-N.J.

The tech industry’s voluntary commitments to the White House on AI are good, but they won’t matter if there isn’t regulation to reinforce those principles, advocates and academics agreed Friday during a Congressional Internet Caucus event. Amazon, Google, Meta, Microsoft, OpenAI, Anthropic and Inflection agreed earlier this month on voluntary measures to ensure AI is developed safely and securely (see 2307210043). Public Knowledge Policy Counsel Nicholas Garcia and Credo AI Global Policy Director Evi Fuelle urged Congress to implement regulations that reinforce the voluntary concepts. Those commitments are going to “fall down a lot” if they remain voluntary, said Fuelle. She said the most important issue is transparency and ensuring the public understands algorithmic modeling and how data is used.

Computer & Communications Industry Association Senior Counsel Joshua Landau warned Congress not to get “too AI-specific” with regulations because existing law can be used to address a lot of the perceived harms. Copyright law and anti-discrimination law, for example, can be applied, he said: Some of the “existential” risks of AI have been “overblown.” Garcia “strongly” agreed existing copyright law can be applied to AI training models. It’s clear that web-scraping is fair use in most instances and shouldn't trigger copyright violations, said Garcia. George Washington University researcher Anna Lenhart highlighted the importance of data scraping in academic work.

Public Knowledge is a longtime advocate for establishing a tech-specific regulator, said Garcia, who supported a bill that would create such a regulator -- the Digital Platform Commission Act from Sens. Michael Bennet, D-Colo., and Peter Welch, D-Vt. The U.S. needs expert regulators with the flexibility to continue to adapt regulations to keep pace with rapidly advancing technology, said Garcia. Lenhart noted the Digital Competition Protection Commission Act from Sens. Lindsey Graham, R-S.C., and Elizabeth Warren, D-Mass., is the first bill to envision an AI licensing regime. Garcia noted the usefulness of FDA licensing requirements for pharmaceutical companies. As did Landau, but he warned that licensing can slow down or impede technological innovation.