Daily Brief SecondSource Morning Brief · October 9, 2026 · Oct 9, 2026
For readers who want to dig in. If you only want what happened, go back to Today.
This issue draws on our internal research brief from early October 9 and last night's sweep. All four main-line items have event dates of October 7 or 8; items 1 and 4 also cite Epoch AI's October 8 monthly brief (its revenue estimate was published October 6 and its cost report September 22); the material in the later columns dates from September 18 to October 8. We swept 450 pieces overnight, and this issue cites 25 outside sources with links you can check.
This is the only census of its kind we found; we read the free portion in full and not what sits behind the paywall. SemiAnalysis itself says "not found" covers only the material it searched, which doesn't mean no testing was done; each model size and snapshot counts as a separate release, so the ratios are indicative only and can't be read as a ranking of companies. We haven't read the TC260 document directly. Epoch's six-company figure isn't audited, and the conglomerates' AI revenue is an estimate. ⚠️ "Pacing" was proposed by Anthropic's chief executive, and our analysis was produced with help from Anthropic's models.
The firing itself has two sources, OpenAI's statement and the researchers' own accounts; we read the three posts and The Star's reprint of The Wall Street Journal's report, not the Journal's original, and the letter is an image we didn't parse. As to why they were fired, the two sides' accounts conflict and neither side has shown public evidence: OpenAI gave no details of the violation and didn't respond that day; Korbak's account is his side alone, and "the real reason was the warnings" is his conjecture. ⚠️ OpenAI competes with Anthropic, and Anthropic's models helped produce this analysis.
The change log is OpenAI's own file, which we read directly. Two counts don't reconcile, and the log doesn't explain either: Roberts' 19 modifications against the log's 14 revised proofs and 13 updated papers, and the 722 main results at release against the log's 719. The 42% counts main results, not every theorem; the 3 withdrawals count papers, not main results; and a withdrawal only reflects an error already found, so for the other roughly 58%, the 419 unformalized main results, there is no error rate yet. Glazer's check is a first pass by a different AI model, not human review or formalization; OpenAI hasn't responded, so it can't count as confirmed.
We checked the composite index, usage and cost directly on the Artificial Analysis page; the page carries no date, and the numbers shift with each rerun. The 160,000 per question and the threefold figure are AINews's arithmetic, not the evaluator's own text. The cost multiple of 2.7 is smaller than the usage multiple of 3, and the source doesn't break down why. One evaluator. ⚠️ Haiku 5.5 is Anthropic's product, and we used Anthropic's models in this analysis.
What you are not getting today. What would most affect judgment is the six-month checkpoints SemiAnalysis itself lists in the paywalled second half, which we haven't read. Beyond that: the TC260 Framework 3.0 text, the full letter from the three researchers to OpenAI's leadership, the details of the violation OpenAI alleges, The Wall Street Journal's original, CrowdStrike's original report, the Lancet paper itself, the measurement date of the Artificial Analysis page, Gemini agent's pricing and timeline, and Cyber Mission's budget and slots; we read none of them. Last night's unread social posts, academic papers and show transcripts went unread because we didn't get to them, not because they couldn't be fetched.
The past 24 hours. 450 new pieces came in from October 8 to 9: 204 academic papers, 143 social-platform posts, 50 other papers (not among the 204 academic papers), 35 company and personal blog posts, 9 subscription newsletters, 7 show transcripts and 2 company filings. Of those 450, we read 25, screened out 22 as off-topic, refetched 1 and did not get to 402. Of the 143 social posts, we read 20, screened out 11, refetched 1 and left 111 unread. Of the 204 academic papers, we screened out 9 and left 195 unread. The 25 we read were 20 social posts, 3 newsletters, 1 blog post and 1 filing; main-line items 1 and 4 come from the 3 newsletters we read (SemiAnalysis, the Epoch brief and AINews), while items 2 and 3, the "Also happened" section and the columns rest mainly on last night's separate pass through social posts and on company pages read directly. To check, we separately read the Artificial Analysis comparison page (tool excerpt), the OpenAI math repository's change log, The Star's reprint, the GlobalFoundries press release, the full text of TSMC's 6-K, the Ministry of Finance statistics page, Google's official blog and Anthropic's official page.
One-time backfill. No new one-time backfill today.
A note on source concentration. ⚠️ Item 1's China figures rest mainly on SemiAnalysis, with Epoch filling in the economics; we read neither the TC260 text nor any Chinese first-hand material. ⚠️ On the reasons in item 2, there is only each side's own account. ⚠️ All three chips items carry Dan Nystedt's relay; the first two also have official documents, the third has only the relay. ⚠️ Our analysis was produced with help from Anthropic's models: item 1's pacing proposal comes from Anthropic's CEO, item 4 and the second Product moves item concern Anthropic products, OpenAI, the subject of items 2 and 3, competes with Anthropic, and the models tested in the second Model watch item include Claude Fable 5.
The sources we track. Our long-term roster has 529 named sources: 302 on social platforms, 90 shows, 51 news outlets, 48 blogs, 48 paper authors and 46 newsletters, plus earnings calls, keynotes and other channels; one source can be counted in more than one channel, so these counts add up to more than 529. ⚠️ Those are counts of tracked sources, a different population from the 450 new pieces above; they can't be added or compared. Representative names: on social platforms, Dan Nystedt, Neel Nanda, Daniel Kokotajlo, Vitalik Buterin and Teortaxes; among newsletters, SemiAnalysis, Epoch AI, Latent Space and Zvi Mowshowitz; among shows, Unsupervised Learning, Cognitive Revolution and ChinaTalk. This issue uses 25 outside sources in the body. We count only links the body actually cites that are not on our own domain, so it can't be compared with the 529 tracked sources either. Of our Taiwanese sources, 7 were added in September for a total of 124; Taiwanese media, Taiwanese companies' official channels and local Taiwanese reporters count, foreign outlets relaying Taiwanese media do not.