Meta Dodges a Bullet, But Still Under Fire in AI Copyright Case
Week of Feb. 24: Amazon Partners with Suno, Billionaires Battle the EU, and More
Meta had lately suffered a series of embarrassing procedural setbacks in the copyright infringement case brought against it by a group of authors over the apparent unlicensed use of their works to train Meta’s Llama Large Language Model (LLM). First came the unsealing of a tranche of previously undisclosed internal communications revealing a discussion among Meta executives over the ethics and legality of its use of BitTorrent networks to download libraries of works it knew to be pirated while also “seeding” those networks with copyrighted works. That was followed by the belated discovery of 18,000 documents it should already have turned over to the plaintiffs. Then came what might have been a coup de grâce when the judge in the case ordered Meta to turn over a group of privileged attorney-client communications to review whether the privilege was concealing the attorney’s participation in a crime.
On Thursday, however, Meta dodged that potentially fatal bullet when the judge ruled the documents contained no evidence of a crime.
“I’ll take this opportunity to say that I’ve reviewed all of the documents submitted in camera, and I did not see any evidence of Meta using its lawyers to facilitate a crime. So the crime-fraud issue is over,” U.S. District Judge Vince Chhabria ruled at a hearing.
Better still for Meta, the judge sharply admonished plaintiffs’ lawyer for the over-heated rhetoric he used in raising the possibility of criminal conduct in the first place.
“The rhetoric that you’ve been using in these hearings and in the letters are, on a scale of one to ten, they are an eleven. You need to dial it back to like a three, because you’re going to lose your credibility very quick, to the extent you haven’t already,” he told attorney Maxwell Pritt.
Meta can’t leave the foxhole just yet, however. The judge rejected for now its motion to dismiss the charge of removing copyright management information from works in the Llama training dataset. And those internal messages about downloading and redistributing pirated works are still in the record. Whether the prove dispositive, or not, they expose the grotesquerie of arrogance, indifference and outright disdain for inconvenient rules at the heart of Meta’s entire AI project — the same appalling mix of character flaws that have lately seemed to recommend Mark Zuckerberg to the mendacious felon in the White House.
Amazon Music Goes AI
Speaking of arrogant billionaires, fresh off destroying the independence and relevance of the Washington Post, Jeff Bezos’ Amazon has partnered with AI music generator Suno to add an AI music slop feature to an updated version of Alexa. According to an Amazon blog post, “Using Alexa’s integration with Suno, you can turn simple, creative requests into complete songs, including vocals, lyrics, and instrumentation.” Bangers all, no doubt.
Suno, of course, has been sued by the major record companies and German collecting society GEMA for copyright infringement. In a court filing last August, Suno acknowledge its “training data includes essentially all music files of reasonable quality that are accessible on the open Internet.”
See No Evil
And still more billionaire flexing: Fresh off his made-for-TrumpTV Oval Office news conference to announce OpenAI’s plans to dump $500 billion into new datacenters, CEO Sam Altman now wants the administrations help to get the European Union to roll back several provisions of the AI Act. Topping the list: the requirement to disclose the contents of the datasets used to train the biggest models. In a Feb. 8th op-ed in Le Monde, Altman wrote, “European regulators, who are working on the application of the AI Act, must think about the consequences of their decisions on tomorrow’s opportunities, especially at a time when the rest of the world is advancing.” One of the consequences, of course, is that disclosure could provide copyright holders with a cause of action against OpenAI and its peers if their works were used in training.
Altman’s comments have been echoed by Google and Meta execs in various forums. “I think there is now broad consensus that European regulation around technology has its issues, and sometimes it’s too fragmented, like GDPR [General Data Protection Regulation], sometimes it goes too far, like the AI Act,” Meta’s director of public policy Chris Yiu told the Techarena conference in Stockholm, Sweden, last week. “But the net result of all of that is that products get delayed or get watered down and European citizens and consumers suffer,” he said.
Trump’s new BFF Zuckerberg, meanwhile, wants the White House to get the EU to stop already with the fining of Meta over its violations of EU privacy, data handling and other laws.