The House Science, Space, & Technology Committee met today to markup nine AI-related bills, approving all of them by bipartisan voice votes. The Center for AI Policy (CAIP) welcomes the progress of these bills through the House and thanks Chairman Frank Lucas (R-OK) and Ranking Member Zoe Lofgren (D-CA) for their efficient work.
In particular, CAIP strongly endorses HR 9194, the Nucleic Acid Screening for Biosecurity Act, and HR 9497, the AI Advancement and Reliability Act.
The Nucleic Acid Screening for Biosecurity Act will push stakeholders toward developing a workable set of standards for screening customers to prevent terrorists and criminals from buying AI-designed DNA. These standards are vitally needed because bio-risk is the deadliest near-term threat posed by AI. It's much harder to stop the spread of an artificially-created virus than to prevent people from creating these bioweapons. Although the standards set by HR 9194 would be voluntary, CAIP is optimistic that a combination of self-interest and pressure from federal grantmakers will motivate the vast majority of biology labs to comply. CAIP applauds Rep. Caraveo (D-CO) for sponsoring this thoughtful bill.
The AI Advancement and Reliability Act will establish the Center for AI Advancement and Reliability, which will have a dedicated mission to promote the measurement and evaluation of AI safety. CAIP believes that current safety benchmarks are already much better than nothing and that companies who are training billion-dollar models should be required to use them. However, there is significant room for improvement, and forming a public-private partnership to design new and improved benchmarks will remove a common objection to mandatory safety audits. CAIP thanks Rep. Obernolte (R-CA) for sponsoring this useful bill.
CAIP is also pleased to see the approval of:
CAIP calls on House leadership to promptly bring these bills to the floor for a vote so that they can be signed into law this year.
These bills are genuinely useful, but they are only the beginning of the massive effort that Congress will need to put forward in order to keep the American public safe in the age of advanced AI.
We need these bills passed now so that next year Congress can move on to more ambitious AI policies.
CAIP supports these reporting requirements and urges Congress to explicitly authorize them
Response to the initial public draft of NIST's guidelines on misuse risk
With the Senate returning today from its August recess, there are two strong bills that are ready for action and that would make AI safer if passed