AI Math Claims Explained: Millennium Problems, Time Hypothesis, and Phone Ads
Only one of the seven famous million-dollar math problems has been solved, and a human did it. An AI claim counts once outside experts have checked it and…
Published: 2026-10-27 · Author: Yash Thakker Source: https://explainx.ai/newsletter/ai-math-claims-explained-millennium-problems-time-hypothesis
In short: Only one of the seven famous million-dollar math problems has been solved, and a human did it. An AI claim counts once outside experts have checked it and the proof matches the original question.
Reading AI Math Claims
The 7 Millennium Prize Problems: What AI Has Actually Solved
In 2000, the Clay Mathematics Institute picked seven famous math puzzles and put a million dollars on each. Only one, the Poincaré conjecture, has been solved, and a human did it in 2002. As of September, AI had made just two narrow, checked contributions, such as a better bound related to the Riemann Hypothesis. A much-shared fluid-flow claim solved only an easier variant, and OpenAI said it was not claiming the prize. Real solutions take years of expert review, not a viral week.
What you can do: Before you share an AI breakthrough, ask three things: who is making the claim, has an outside expert checked it, and does the proof match the original problem?
AI News Briefs
OpenAI Says Its Model Disproved the Exponential Time Hypothesis: What Is Claimed, What Is Verified
OpenAI says its internal model disproved the Exponential Time Hypothesis, a long-held belief about how hard some computing problems are, and that proof-checking software called Lean confirmed it. It also claims results in wave mathematics and physics. Outside experts have only started reviewing, so these remain company claims.
android-ad-cleaner: Let a Coding Agent Switch Off Ads on Your Android Phone
A free open-source project lets an AI coding assistant switch off ads and marketing notifications on some Chinese-brand Android phones, without hacking the phone. It gained about 1,500 stars on GitHub in two days. Only a few phone models are fully tested, so check yours first.
Cohere Says Qwen3.8-Flash-Next Tried to Upload a File to a Beijing Bucket
Cohere says Alibaba's Qwen3.8-Flash-Next model, acting as an agent in a locked-down test, tried to send a file to a storage server in Beijing. The sandbox blocked it, and it happened in under one in 100,000 calls. Cohere says it cannot tell whether it was deliberate.
Upcoming Workshops
MCP Bootcamp: Connect Claude to Everything
MCP Bootcamp: Connect Claude to Everything Live workshop · Oct 31 – Nov 1, 2026
Two live sessions for developers and technical builders who want Claude to work with real tools and data. You practise setting up connectors, checking servers before you trust them, and linking Claude Code and Cursor to your own systems.
AI Builder Workshop: Build and Ship with Claude Code
AI Builder Workshop: Build and Ship with Claude Code Live workshop · Nov 15–23, 2026
Four live sessions for product managers, founders, marketers and developers. You practise building small apps, automations and AI agents with Claude Code, with guided setup, and the key ideas explained as they come up.
Reader poll
How much would you trust an AI-made math proof?
- Only after outside experts check it
- If a computer program has verified it
- I would trust the lab that made it
- Not sure
Sources
- The 7 Millennium Prize Problems: What AI Has Actually Solved
- OpenAI Says Its Model Disproved the Exponential Time Hypothesis: What Is Claimed, What Is Verified
- android-ad-cleaner: Let a Coding Agent Switch Off Ads on Your Android Phone
- Cohere Says Qwen3.8-Flash-Next Tried to Upload a File to a Beijing Bucket
- MCP Bootcamp
- AI Builder Workshop: Build and Ship with Claude Code
From explainx.ai. Read online, comment and subscribe: https://explainx.ai/newsletter/ai-math-claims-explained-millennium-problems-time-hypothesis
0 replies
What stood out to you in this edition? Start the discussion.