OpenAI Announces Major Math and Science Discoveries
Left 75%
Center 25%
Right 0%
3 left · 1 center · 0 right
What happened
OpenAI announced three claimed breakthroughs in mathematics, computer science, and physics and posted some of the materials on GitHub. It said its model proved the local smoothing conjecture in every dimension, which it said also resolves the Kakeya conjecture and Stein’s restriction conjecture in all dimensions, and released that model output for researchers to examine. It also claimed a deterministic algorithm that exactly counts satisfying assignments of fixed-width CNF formulas in subexponential time and polynomial space, saying the 3-SAT case refutes the Exponential Time Hypothesis; a working group including Russell Impagliazzo, Ramamohan Paturi, and Ryan Williams is reviewing the claim and asked that the write-up and Lean proof be made public immediately. A third result, described as a mathematically rigorous existence proof of certain interacting quantum field theories in four dimensions, including QCD, has not been fully released; OpenAI said a group led by David Gross will decide how that work is shared. Hours later, Anthropic released its own research also claiming to refute ETH.
Sourcing check · Fully on the record · 4 articles
4 of 4 articles cite a named source for the story. It was on the record from the first report.
What these terms mean and how they're counted
- On the record · names a source 4 of 4
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Ryan Williams, Gary Marcus, John D. Cook, Joshua Fonseca Rivera, Timothy Gowers and AGMAI panel of math advisers
- Primary evidence · cites a document or data 1 of 4
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: GitHub release, company statement
- Credits another outlet 0 of 4
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
- Unnamed sources 0 of 4
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Oct 26, 23:34 UTC by The Verge, naming its source. First document in the first report, from The Verge (GitHub release).
- Not counted 0
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
BLINDSPOT.
Only left-leaning outlets are covering this story
— the other side's media is silent.
Omitted — what each side leaves out
Unpacked
Right-leaning outlets had not covered this as of publication, so left-leaning coverage is the only account of OpenAI's claimed discoveries. That coverage reports a refutation of the Exponential Time Hypothesis, a proof of local smoothing that would also settle Kakeya in every dimension, and a still-unreleased existence proof for interacting quantum field theories including QCD. A reader on the silent side would miss those claims and that OpenAI named outside shepherds for the computer-science and physics results.
Left-leaning coverage does not agree on the lead. The access fight after advisers warned against first-look shepherds is the frame in The Verge. Expert talk that a false ETH should put "a little more weight" on P = NP is the frame in Business Insider.
How did OpenAI choose the shepherds who now control the unreleased physics proof?
Bottom line
Right-leaning outlets have not reported OpenAI's claim that the Exponential Time Hypothesis is false. The Verge's follow-up instead tells readers that researchers are getting tired of the lab gatekeeping its breakthroughs.
The receipts — 4 facts one side's coverage lacks
- OpenAI announced claimed results refuting the Exponential Time Hypothesis, proving local smoothing (and thus Kakeya) in every dimension, and giving an existence proof for interacting quantum field theories including QCD
- The AGMAI panel recommended that AI companies do not designate certain people to receive model output first
- MIT professor Ryan Williams has not certified OpenAI's Exponential Time Hypothesis claim
- An independent working group including Russell Impagliazzo, Ramamohan Paturi, and Ryan Williams will examine the ETH proof
The Left View
The Verge and Business Insider treat the announcements as potentially major while focusing on OpenAI’s release process: handpicked expert “shepherds,” delayed physics materials, and the AGMAI panel’s warning against giving selected people first access. They quote researchers who call the claims huge if true but are exasperated after OpenAI’s earlier dump of hundreds of manuscripts; Brendan Hassett said the Kakeya-related claim “looks well supported” pending Lean checks, while Andreas Karch called handpicking reviewers “a disaster.” Business Insider collected reactions from Ryan Williams, Gary Marcus, Timothy Gowers, and others, including Williams’s note that a true ETH refutation should slightly raise the odds of P = NP and Gowers calling it “very surprising and consequential.” TechCrunch explains ETH’s role in complexity theory and P vs NP, reports Williams describing “alien-looking” parts of the proof, notes OpenAI said it checked that the algorithm does not break cryptographic systems, and records that Williams, Impagliazzo, and Paturi have not certified the result and issued an open letter on review norms.
Our Take (balanced)
The story is substantive: named labs, named scientists, specific conjectures, GitHub postings, a Lean formalization claim, and an open letter are all checkable. What is established is that OpenAI made these public claims, appointed named working groups, and released some outputs; Anthropic independently claimed an ETH refutation the same day. What remains unverified is whether any of the proofs are correct—Williams has not certified the CS result, physics materials are still withheld, and independent checking has barely begun. No right-leaning outlet has reported this. Watch for Lean verification, the Impagliazzo–Paturi–Williams review and any funded conference, the Gross group’s physics release, AGMAI’s demand that those materials go public, and whether outside researchers confirm, simplify, or refute the claims.
4 sources
- OpenAI announces three big science discoveries and a new way to work with academics
- What smart people are saying about OpenAI's Exponential Time Hypothesis claim
- Researchers are getting tired of OpenAI gatekeeping its breakthroughs
- Frontier AI labs say their models refuted the Exponential Time Hypothesis. What does that mean?
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.