Anthropic's Claude solves nine-loop amplitude challenge in physics
Summary
In September 2026, researchers from Anthropic announced that their AI model, Claude, successfully computed the nine-loop MHV six-particle amplitude in planar N=4 super Yang-Mills, marking a significant advancement in theoretical physics. This achievement surpassed the previous limit of eight loops set by physicist Lance Dixon and collaborators, who utilized the same model as a testing ground for particle interaction predictions. Claude's task, approached autonomously with minimal human input, leveraged established computational methods and cost only a few thousand dollars, highlighting the potential of AI in solving complex scientific problems that have historically required substantial computational resources and expert effort.
Analysis
Claude: Claude is Anthropic's family of large language models designed for advanced reasoning and task execution. In this news, Claude autonomously performed a complex bootstrap calculation to solve the nine-loop hexagon amplitude in planar N=4 super Yang-Mills theory, following methods from human physicists and operating within a structured scientific harness. Anthropic: Anthropic is an AI research company focused on developing safe and capable large language models, including the Claude series. Here, physicists on its team issued a targeted prompt to Claude, oversaw its multi-day unsupervised run, and confirmed the result with independent verification by external experts. Lance Dixon: Lance Dixon is a professor of particle physics and astrophysics at SLAC National Accelerator Laboratory and Stanford University, specializing in scattering amplitudes. He independently validated Claude's nine-loop result using established techniques from his prior work on related amplitudes and form factors. Matt von Hippel: Matt von Hippel is a theoretical physicist turned science writer and blogger at 4gravitons.com. He formulated the original challenge for AI systems to advance loop-order calculations in toy models of particle physics, which directly prompted Anthropic's effort described in the news. Expert Validation: Results from AI-assisted computations in amplitudes are being cross-checked and extended by leading human physicists using established mathematical frameworks. AI Scientific Workflows: Anthropic has expanded Claude into dedicated platforms that enable autonomous handling of multi-step computational research tasks with minimal human intervention. Physics AI Applications: Large language models are now being applied to frontier problems in theoretical particle physics, including amplitude calculations that previously required extensive expert effort.
Categories
machine_learningaitechai_agents
Related sources
- https://quantumzeitgeist.com/anthropic-ai-capacity-ais-frontier/
- https://openai.com/index/navier-stokes-solution/
- https://openai.com/index/new-result-theoretical-physics/
- https://arxiv.org/abs/2609.19252
- https://tech-insider.org/navier-stokes-solved-2027-manifold-odds-claude-rumor/
- https://www.nature.com/articles/s41567-026-03436-8
- https://github.com/consensusframework/yang-mills-mass-gap
- https://thenextweb.com/news/anthropic-claude-science-ai-workbench-scientists
- https://www.nytimes.com/2026/09/24/science/anthropic-biology-lab-enzyme.html
- https://www.reuters.com/science/anthropic-unveils-claude-science-ai-platform-scientific-research-2026-06-30/
- https://www.scientificamerican.com/article/anthropics-ai-steals-mathematicians-record-for-most-complicated-curve/
- https://techcrunch.com/2026/06/30/anthropics-claude-science-bets-on-workflow-not-a-new-model-to-win-over-scientists/
- https://www.science.org/content/article/chatgpt-spits-out-surprising-insight-particle-physics
- https://arxiv.org/abs/2607.21629
- https://www.technologyreview.com/2026/06/30/1139987/claude-science-is-anthropics-newest-flagship-product/