AI’s Proof Problem: Who Checks the Breakthroughs?

OpenAI says an internal model has resolved more than 100 open maths problems, while a new independent advisory group considers how such results should be reviewed and communicated. We examine the unverified claim, the group’s limited authority and the strain AI-generated research could place on scholarly validation. Also: proposed international standards for automated AI research, a coalition tackling the language gap, NVIDIA’s power and cooling qualification program, Amazon’s block on Meta’s Muse shopping agent, a proposed US-China AI incident alert channel, GitHub’s enterprise credential inventory, and what Googlebook changes for laptop buyers and developers. Full transcript: https://www.owenonthenet.com/podcast/2026-09-22/ Sources: https://openai.com/index/advisory-group-on-mathematics-and-ai https://terrytao.wordpress.com/2026/09/21/advisory-group-on-mathematics-and-artificial-intelligence https://openai.com/index/building-standards-next-phase-ai https://www.gatesfoundation.org/ideas/media-center/press-releases/2026/09/ai-language-partnership https://apnews.com/article/artificial-intelligence-anthropic-openai-gates-aefb021bede3b02c83890f65cd540fd0 https://blogs.nvidia.com/blog/dsx-ready-ai-factories-power-cooling https://www.geekwire.com/2026/amazon-blocks-metas-muse-ai-assistant-in-new-standoff-over-agentic-shopping https://apnews.com/article/bessent-ai-xi-trump-china-trade-2c7f54f07e755f506d9db9b91df282bd https://github.blog/changelog/2026-09-21-github-enterprise-adds-credential-inventory-exports https://blog.google/products-and-platforms/devices/googlebook/googlebook-built-in-intelligence