Anthropic has reclaimed the spotlight in the artificial intelligence sector after its Fable 5 model successfully disproved a mathematical conjecture that had remained unsolved for 87 years. The model produced a counterexample to the Jacobian conjecture, a central open problem in algebraic geometry originally posed in 1939. This achievement underscores the potential of AI to contribute to complex scientific discovery, a development with significant implications for the technology's future applications and valuation.

Anthropic researcher Levent Alpoge announced the finding on social media platform X on Sunday, noting the counterexample is just 216 characters long. The simplicity of the solution allowed mathematicians to verify it quickly, with Abhishek Saha, a mathematician at Queen Mary University of London, describing it as "a pretty big deal." He suggested it is likely the most significant conjecture AI has helped settle to date.

Chinese Labs Intensify Competition

The mathematical breakthrough capped a week of aggressive activity in the AI sector, particularly from Chinese frontier labs. Moonshot AI released its Kimi K3 model on Thursday, featuring 2.8 trillion parameters. The launch briefly impacted U.S. chip stocks and put pressure on domestic competitors; shares of Zhipu and MiniMax both fell following the announcement.

Demand for Kimi K3 was intense enough that Moonshot AI paused new user signups within two days. Analysts suggest this compute constraint could benefit hardware providers like Nvidia Corporation and Micron Technology Inc. as companies scramble to secure infrastructure for training and running large models.

Market Leaders and Public Proxies

Alibaba Group Holding previewed its Qwen3.8 Max model on Sunday, claiming it is "second only" to Anthropic's Fable 5, though the company did not publish benchmarks or independent tests to support the assertion. Despite the lack of verification, Alibaba shares rose by as much as 5.4% on Monday. Polymarket traders currently give Alibaba a 79% chance of having the best Chinese AI model by the end of July, significantly ahead of Moonshot AI at 16%.

On a global scale, Anthropic is viewed as the clear leader. Traders on Polymarket assign a 99% probability to Anthropic possessing the best AI model by the end of July, with volume exceeding $7 million. Looking further ahead, Anthropic holds a 65% chance of leading by year-end, followed by Alphabet Inc. at 13%. Moonshot AI remains the top Chinese contender in the global race at 3%.

Anthropic may transition from a private to a public entity in the near future. Reports indicate the company filed a confidential S-1 in June following a funding round that valued the company at $965 billion. Polymarket traders place the odds of an initial public offering by year-end at 64%. Until a listing occurs, backers Amazon.com Inc. and Alphabet Inc. remain the closest public proxies for investors seeking exposure to Anthropic's success.

A federal judge in San Francisco granted final approval to Anthropic's $1.5 billion class-action settlement with a group of authors who accused the artificial intelligence startup of improperly using their books to train its Claude chatbot. U.S. District Judge Araceli Martinez-Olguin signed off on the agreement, which is believed to be the largest copyright settlement in U.S. history, rejecting objections from authors who argued the deal was inadequate.

The lawsuit, originally filed in 2024, is one of dozens brought by authors, publishers, and other copyright holders against AI companies over the use of copyrighted works to train large language models. It is also the first major AI copyright case in the U.S. to settle. The agreement had received preliminary approval last year from now-retired U.S. District Judge William Alsup.

Legal Rulings and Fair Use

Last year, Alsup ruled that Anthropic's use of books to train Claude qualified as fair use under U.S. copyright law. However, he also found that the company violated copyright by maintaining a digital repository of more than seven million pirated books that were not necessarily used for AI training. A trial to determine damages related to those pirated copies had been scheduled for last December, with potential liability that could have reached hundreds of billions of dollars before the parties agreed to settle.

“We reached this settlement in 2025, after the court’s landmark ruling that training AI on books is fair ​use under copyright law — which remains the law today,” Anthropic Deputy General Counsel Aparna Sridhar said in a statement. She added that more than 91% of eligible authors and publishers have already claimed their share of the settlement proceeds.

Settlement Details and Attorney Fees

Some authors challenged the settlement, arguing it undervalued their claims, overcompensated plaintiffs’ attorneys, and excluded certain copyright holders. Martinez-Olguin rejected those objections, stating they were “not grounded in ⁠a realistic ​assessment of the overall risks and rewards of a trial.”

The judge awarded plaintiffs’ attorneys more than $101 million in legal fees, a portion of the $187.5 million they had requested. “It ​is the largest known copyright recovery in history. We look forward to making distributions to the Class as promptly as possible,” lead plaintiffs’ attorney Justin Nelson said in a statement. Some authors and publishers opted out of the class action and continue to pursue separate copyright lawsuits against Anthropic.