Judiciary Panel Debates AI, Deepfakes as Shutdown Continues

The ongoing government shutdown’s impact on the federal courts was front-and-center at a judiciary panel meeting, where judges and lawyers debated ways to address artificial evidence and so-called deepfakes despite furloughs of court staff.

Judge Jesse Furman, chair of the Judicial Conference’s advisory committee on evidence rules, held the meeting Wednesday to consider various proposals to amend court rules from New York, with most attendees appearing virtually, rather than in-person in New Orleans as originally planned.

On several occasions throughout the roughly three-hour meeting, attendees noted the committee’s progress was hindered in part by the funding lapse, which led to shutdown-related furloughs of court employees last month for the first time in decades. The judiciary ran out of money in mid-October amid congressional stalemate to fund the government, and judiciary employees will see their first missed paycheck Friday if the shutdown continues.

Furman, of the US District Court for the Southern District of New York, said the panel could work with the Federal Judicial Center, the judiciary’s research arm, to send out a survey that would collect information on how often deepfake evidence is a problem in courts, after several judges expressed interest in learning more before proposing a change.

But he acknowledged the panel would have to wait for the FJC, affected by furloughs, to begin fully operating again.

Betsy Shapiro, a Justice Department lawyer on the committee, said she’d been “handicapped” by the shutdown in her ability to gather more information about the impact on government software of a proposed change to how machine-generated evidence is examined, as a number of people who would otherwise contribute to her preparation have been furloughed.

In considering a separate proposal that would affect Native American tribes, Shapiro said she’d circulated materials to the Justice Department’s Office of Tribal Justice, but got furlough responses from every person.

Furman thanked committee staffers multiple times during the meeting for working under “extraordinary circumstances” and through “hardship.”

The panel’s next meeting will be held in Washington on April 28, Furman said. “God help us if the government is still shut down or shut down again,” Furman said.

AI Changes

The evidence rules panel was weighing a proposal, posted earlier this year for public feedback, to mandate that “machine-generated evidence” must be subject to existing requirements on expert witness testimony, with an exception for output by “simple scientific instruments.”

Committee members raised concerns about the sweep of both of those terms and discussed alternatives, including computer-generated and machine learning.

Andrea Roth, a professor at University of California, Berkeley Law, who proposed some changes to the committee, warned that the use of the words “simple” and “scientific” could “create mischief and unnecessary litigation.”

She also said the proposed phrase “machine-learning” was “both under and over inclusive,” as it may omit complex algorithms that merit scrutiny.

The proposal was posted for public comment in June, and is open for comments through mid-February.

Furman said the proposal hasn’t received any comments yet, but that he’s hopeful it draws “a rich body of comments” before the period closes. He also noted that its release “does not mean that we are committed to adopting this rule.”