An artificial intelligence bill championed by Senate Commerce Chair Maria Cantwell (D-WA) that cleared committee July 31 includes an amendment from ranking member Ted Cruz (R-TX) that observers say would neutralize key aspects of President Biden’s AI executive order while a committee aide said the chair will work to remove the language as the bill moves through remaining barriers to final passage.
“This Cruz amendment doesn't just destroy the bill that Cantwell is trying to pass. It also guts the AI EO, the [Office of Management and Budget] guidance and any future action by the federal govt on AI. And it passed the committee!!!” reads a post Suresh Venkatasubramanian shared July 31 on X with a link to the amendment.
Venkatasubramanian directs the Center for Technological Responsibility, Reimagination, and Redesign at Brown University’s Data Science Institute and coauthored the White House’s AI Bill of Rights, according to his bio on the social media platform.
The committee late on July 31 posted links to the various amendments added to legislation that passed in a massive markup earlier in the day.
The Future of AI Innovation Act -- which the committee highlighted in a press release following the session and was itself a substitute amendment from the chair -- for example, included nine additional amendments.
Amendment 4 by Cruz, which Venkatasubramanian flagged, would require the president, within seven days of the bill’s enactment, to issue a directive that would, among other things, prohibit federal agencies from requiring or recommending “artificial intelligence, algorithms, or other automated systems [to be] designed in an equitable way that prevents disparate impacts based on a protected class or other societal classification.”
Critics say that runs counter to the March 28 memorandum OMB issued for implementing Biden’s landmark Oct. 30 EO on AI, which instructs agencies to ensure their use of AI protects individuals’ safety and rights.
A calculated move by the chair
A spokesperson for the committee confirmed passage of the amendment saying it was part of a calculated move by Cantwell.
“Yesterday’s markup is just the first step in advancing legislation through the Senate. Chair Cantwell is committed to protecting people from harm caused by artificial intelligence, including bias and discrimination,” the spokesperson said, noting, “Sen. Cruz has made clear -- through his public statements and the dozens of divisive, radical amendments filed to bipartisan AI legislation -- that he is not for making sure AI is developed in a way to protect consumers from harms, ensure their safety or defend their constitutional protections.”
The committee spokesperson said: “Rather than giving the senator a platform to amplify divisive rhetoric and delay committee progress, the Chair accepted the amendment—knowing there will be many opportunities to fix the legislation, and with the support of the other Republican Senators.”
But the move stunned seasoned observers in the policy space.
“The Senate Commerce Committee just approved this terrible provision on AI from Ted Cruz???” wrote Nik Marda, also observing passage of the amendment on X. Marda is the technical lead for AI governance at Mozilla but was only commenting in his personal capacity. He is also a former chief of staff for technology at the White House Office of Science and Technology Policy.
One AI policy expert who spoke with Inside AI Policy but declined to be identified as they awaited confirmation that the amendment had indeed passed, said the amendment goes far beyond undoing the Biden EO.
“We want to confirm what the current status of that bill is and what the intent is going forward for its sponsor. It’s a very severe amendment, it would be very problematic,” the person said, adding, “There’s two sets of threats here. The first is kneecapping the administration’s AI policy efforts generally, and the second is that the language around disparate impact and things like that in the Cruz amendment threatens the enforcement of civil rights law across the entire federal government, across numerous pre-existing legal regimes.”
Disagreement over the need to conduct impact assessments given the tendency of AI to replicate harmful bias was also in play after House Energy and Commerce Chair Cathy McMorris Rodgers (R-WA) removed a substantial section on algorithmic accountability from a privacy bill she originally drafted with Cantwell, following industry opposition. Civil society groups said the move was “unacceptable.”