The ruling dismissed the majority of claims in the legal action initiated by writers who claimed their works were utilized, without consent, to train Anthropic’s Claude AI systems. Just a single claim — related to Anthropic’s early use of stolen copies for its main library — was permitted to continue. The remainder were struck down. The company will still have to address, through litigation, its training methods involving more than 7 million illegally obtained titles.

On the whole, this is seen as a success for Anthropic. However, it comes with uncertainty regarding copyright law.

Expert Reaction To AI Copyright Decision — ‘This is bad law’

Yelena Ambartsumian holds nothing back.

“If you’ve reviewed the Bartz v. Anthropic decision on copyright violation, you’ll notice that the order sets a poor precedent,” stated Ambartsumian, intellectual property lawyer and founder of Ambart Law, PLLC. “It's evident Judge Alsup aimed to safeguard and encourage generative AI innovation, and that’s precisely what he achieved.”

Her perspective? The verdict overlooks essential copyright principles. Most notably, she pointed out, is the judge’s reasoning about reproductions. The Copyright Act grants authors sole rights to reproduce their work. This includes the kind of extensive duplication involved in AI training. Yet Judge Alsup ruled that no infringement occurs “where one copy completely supersedes the other.”

Ambartsumian is direct. “According to this decision, if I scan a book to make a digital version, and then discard the original, there’s no breach of copyright, because my digital version took its place. That can't be correct.”

Fair Use For AI Gets Distorted

Judge Alsup also heavily relied on fair use. But the analysis was shallow where it mattered most.

Rather than concentrate on the commercial application of protected material, the judgment likened AI training to how people learn. That’s a precarious comparison since humans don’t profit from their mental processes at industrial scale.

“This is a revenue-generating machine, not a human learner,” Ambartsumian noted. She concedes the analogy may work conceptually, but insists that doesn’t mean the legal handling should follow suit. “I've previously written that I believe copyright law might not be the ideal path to tackle issues of uncompensated copying,” she continued. “But this rationale skips over the most crucial components of the fair use evaluation and fails to consider the commercial nature involved.”

She also highlighted that Judge Alsup overlooked the magnitude of what was copied. Anthropic didn’t just take excerpts. It consumed entire books. Rather than confronting that directly, the court concentrated on whether Claude required full books or only partial content. The verdict? Using everything accessible enhances the model, so it’s acceptable.

“That’s exactly what I warned against,” Ambartsumian said, alluding to her response to the Copyright Office’s prior report on AI training. “The Copyright Office suggested that training on massive amounts of varied content could enhance a model's robustness and therefore more likely be considered ‘transformative.’ This encourages even more copying, rather than less, and might actually promote copyright violations.”

Judicial Discretion And Selective AI Model Favoritism

Another factor lurks beneath the surface. Anthropic may have gained favor due to being perceived as the “more responsible player” in the AI space.

“Would OpenAI, which some might view as the less scrupulous elder sibling of Anthropic, have received the same treatment from the same judge? Unlikely,” Ambartsumian remarked. “In copyright cases, we frequently observe judges bending existing rulings to support the ‘good actors’ while punishing ‘bad ones’ — or ignoring creators altogether.”

That inconsistency now extends into courtrooms. In the ongoing New York Times lawsuit against OpenAI, the courtroom atmosphere has been colder, with sharper skepticism.

So where does that leave us? In legal uncertainty. The temporary appeal period remains open. But Ambartsumian isn’t hopeful.

“As a litigator and copyright law enthusiast, I’m disheartened,” she admitted. “I would have preferred to see a stronger explanation for why Anthropic’s training was transformative. But as someone grounded in legal realism, I doubt an appeals court would come to a different conclusion.”

What started as a battle over rights turned into a judgment on priorities. Innovation was chosen. Creators were left with leftovers.

ForbesDecentralized AI Is Watching — And Understanding — EverythingBy Tor Constantino, MBA

The above is the detailed content of Judge Backs AI In Copyright Case — Expert Says It Creates More Questions. For more information, please follow other related articles on the PHP Chinese website!