An unreleased Anthropic AI model has made measurable progress on the Riemann hypothesis, one of mathematics' most famous unsolved problems that has resisted solution for over 150 years.
The Riemann hypothesis, formulated in 1859, concerns the distribution of prime numbers and remains one of seven Millennium Prize Problems. A $1 million reward awaits anyone who can prove or disprove it.
Anthropic’s achievement marks a notable moment in AI's expanding role in mathematical research. While the model did not solve the hypothesis outright, it demonstrated capabilities beyond typical AI applications in this space.
The company has not released detailed technical specifications about which model achieved this progress or the exact nature of the breakthrough. The development underscores the growing intersection between large language models and formal mathematics, an area where AI has traditionally struggled.
Mathematicians have attacked the Riemann hypothesis using various approaches over the past 150+ years. The hypothesis states that all non-trivial zeros of the Riemann zeta function lie on a critical line in the complex plane. Verification of billions of zeros supports the hypothesis, but a complete proof remains elusive.
AI systems like Anthropic's models can process mathematical literature, identify patterns, and generate novel approaches to long-standing problems. However, they operate differently from human mathematicians and sometimes lack the intuitive leaps that characterize breakthrough proofs.
Anthropic’s work joins other recent instances of AI contributing to mathematics and physics. DeepMind's AlphaFold solved protein structure prediction, demonstrating AI's potential in fundamental science. Other organizations have explored AI applications in theorem proving and mathematical discovery.
The unreleased status of Anthropic's model suggests the company is conducting additional validation and refinement before public release. The company typically subjects new models to rigorous testing before announcement.
This development carries implications for how researchers approach unsolved mathematical problems going forward, potentially opening new avenues for investigation alongside traditional human expertise.
SpaceXAI has unveiled new artificial intelligence software designed to operate as a team of coordinated AI agents. The release represents the company's latest effort to compete with leading AI developers Anthropic and OpenAI.
Mark Zuckerberg has revived his push for open-source AI development in a lengthy essay, arriving as Chinese models increasingly match Western performance at significantly lower costs.
Nvidia released Nemotron 3.5 Lightning, an open-weight model with 3.6 billion active parameters that matches larger competitors in intelligence benchmarks while delivering nearly 670 tokens per second, positioning efficiency as a core strategy.
Spotify is introducing 'AI Persona' labels for artificial artist identities and will exclude their music from algorithmic and editorial recommendations by default, rolling out in mid-September.