A rival human-written draft, Microsoft's new Copilot and a Wikipedia compressor: the week's AI stories in plain English.

News · 2026-11-20
Levent Alpöge, a mathematician at Anthropic, says he and Claude found a proof of the Riemann Hypothesis, a 167-year-old problem about prime numbers. He also posted computer-checkable proof code. MIT's Andrew Sutherland reports it passes several independent checks, but the final check was still running and the Clay Mathematics Institute still lists the problem unsolved.
Why it matters: Computer checking makes this claim stronger than most. Experts must still confirm the code states the real problem. It is one person's claim, not an Anthropic announcement.
What you can do: Wait for named experts or the Clay Mathematics Institute before you share it as fact. Write "an Anthropic mathematician claims", not "Anthropic proves".
Read the sourceNews · 2026-11-20
Ravi Vakil, president of the American Mathematical Society, posted Alessandro Masullo's proposed proof of the Generalized Riemann Hypothesis, a broader version of the same problem. He says Brian Conrad reviewed it for months and that no AI was used. Nobody has verified it yet.
Why it matters: Two different routes to the same famous problem arrived within a day. Both are unchecked claims, so neither settles anything yet.
Read the sourceNews · 2026-11-18
At its Ignite event, Microsoft announced a new Copilot, new coding models, a spending view for AI agents and a security system. Many items are generally available, but several agent features are still in early access. Office Watch counted 76 of 215 features as generally available.
Why it matters: Check each feature's status before you plan around it. "Announced" and "available today" are often not the same thing.
Read the sourceNews · 2026-11-18
The Center for AI Safety published the MisAnthropic Statement, signed by scientists, legal scholars and tech leaders. It argues Anthropic markets safety while pursuing self-improving AI and softening earlier safety pledges, as its public share offering began. Our post has no formal reply from Anthropic.
Why it matters: This is advocacy, not a court order. It shows that Anthropic's safety message now faces public scrutiny from named experts.
Read the sourceNews · 2026-11-20
Tufa Labs says its free tool tufazip compresses 1 GB of Wikipedia text to 105.39 MB without losing anything, using one neural network. The public benchmark page confirms the size. But one run takes about 7.5 days on two high-end GPUs.
Why it matters: A model that predicts text well can also compress it well. This is a research result, not a tool for your own files.
Read the source
From us
Two live evenings on Nov 28–29, open to beginners. You build a knowledge base, an agent skill, a prompt library and your own Artifacts, and you connect Claude to the apps you already use.
See the details
From us
Two live evenings on Dec 20–21, for beginner to intermediate learners. You explore Microsoft 365 Copilot, use it in Excel and other Microsoft apps, try Copilot Cowork, and build automations in Copilot Studio.
See the detailsOne vote per browser. An informal reader poll, not a representative survey.
Comments
Join in with your explainx community account. Comments are plain text (no links, up to 600 characters), follow the community guidelines, and are reviewed before they appear.
No comments yet. What did you think of this edition?
Loading…