Appearing before the Advisory Committee on Rules of Evidence in Boston, the two groups concluded that the scientific detection of computer-generated images was doomed to failure. The judges remain cautious about amending Rule 901.
On 15 October in Boston, the American Judicial Conference's Rules of Evidence Advisory Committee heard from OpenAI and Meta about the content generated by IA. Shira Anderson, a public policy lawyer at OpenAI, said artificial images are progressing so fast that there will probably no longer be a scientific method for detecting them. Devin Norder, product manager at Meta, has abounded: the productions of the latest models can be impossible to distinguish from genuine recordings for a human.
A problem still rare in the courtrooms
The magistrates do not rush. An investigation by the Federal Judicial Center showed that only 15 of the 931 judges who responded had already had to decide a challenge to evidence presented as a deepfake. The chair of the committee, Justice Jesse Furman, acknowledged that the announced wave of litigation had not yet occurred, and Justice Valerie Caproni cautioned against AI-specific rules adopted without solid data.
On the table is a draft Rule 901(c): the party who accuses a coin of being rigged should first support its challenge, after which the party who produces it should demonstrate that it is more likely to be genuine than not. The text was not adopted; A subcommittee was mandated to continue the review.
Read the full article on the original source
Go to the source →