Daily Brief Nov 5, 2026
How we checked every item today except "Also happened"
For readers who want to dig in. What happened, in brief, is on the Today tab.
This issue draws on material from November 4 to early November 5; main-line item 1 is also checked against Texas Tribune reports of August 3 and September 21 and a July 23 press release from Hinojosa's campaign. The Named commentary item is an October 31 post, and the From the archive item comes from a SecondSource call of July 14. Last night's collection took in 1,375 pieces, and this issue uses 20 outside sources with checkable links; the later sections also draw on material from the past week.
1. [Today] Scientific American's tally of 41 data-center ballot measures in ten states finds one result in the industry's favor, as Texas picks a governor who backed a pause
Not proven yet： no report says a governor can pause private projects
How we checked
The governor's race result was reported separately by the Texas Tribune and Yahoo Finance; the official count isn't complete. Yahoo Finance's more-than-a-quarter share is one sentence with no statistical source. CNBC reporter Deirdre Bosa wrote on November 2 that Texas is "about a quarter of the US pipeline," a different measure, also unsourced. The count of one industry win among 41 measures is Scientific American's tally alone; other trackers count 38 or 47 measures, and Scientific American doesn't break down how many of the 41 are bans and how many only add review. Abbott's September 21 order was reported only by the Texas Tribune. No company has yet called permits a new constraint on an earnings call or in a filing.
2. [Today] Ben Thompson says politics is the last constraint on AI compute, and scarcity will last far longer than expected
Not proven yet： every point in Thompson's piece is a day-after inference with no numbers
How we checked
The "more than 400" comes from Xu's own dashboard, whose method isn't public; another tracker, GatherGov, counted 632 jurisdictions in September, including measures still under discussion. The congressional results Thompson cites are Decision Desk HQ projections as relayed by Bloomberg, not final counts.
3. [Today] OpenAI says its newer cybersecurity model found 948 Linux kernel flaws that grant root, against 166 for the model it used in July
Not proven yet： every figure is self-reported by OpenAI; no third party has re-run it
How we checked
All figures are OpenAI's measurements in OpenAI's own environment. The original text calls the model used in October only "a newer cybersecurity model," and another section compares GPT-5.5 Cyber with GPT-5.6 Cyber. The original gives only the ten-day figures and does not say which reports maintainers will eventually accept.
[This week] (October 31) Exponential View author Azeem Azhar pushes back on McKinsey: what data centers really compete on is how fast they get power, not the power price.
How we checked
Neither side attached cost-structure figures, and McKinsey doesn't say how the US$200 and US$380 were calculated.
(McKinsey Global Institute, 2026-10-31) (Azeem Azhar, 2026-10-31)
[This week] (paper, November 3) Bloomberg self-reports that AI-assisted migration across thousands of internal codebases opened 2,333 merge requests, the code changes sent to a team for review, of which 729 succeeded.
How we checked
[Today] Anthropic's biology team says that with Claude involved at every step, the team solved the first structure of a prokaryotic vault particle in four months and cut one purification step from three days to 90 minutes.
How we checked
The time savings are the team's estimates, and the preprint has not been peer reviewed. Vault particles are large hollow protein shells, present in almost every human cell, whose function is still unknown. Purification fell from three days to about 90 minutes, and the first structural model took 34 hours, against what the team estimates would have been weeks.
[Today] Databricks says Unity Gateway customers used an average of 7.1 models in July, up from 4.2 in January.
How we checked
Databricks gives that figure while promoting its gateway and gives no January open-weight share to compare with.
[Today] Perplexity's Decisions API can now search the web before answering, and on Perplexity's own test true/false accuracy rose from 56.4% without search to 97.3% with it.
How we checked
[Looking back] (SecondSource call, July 14) Enterprises can teach a general model their business through retrieval or fine-tuning; SecondSource's reading is that the two split the work rather than one replacing the other.
How we checked
Sources & accounting
The past 24 hours. 1,375 new pieces came in from November 4 to November 5: 929 academic papers, 160 linked articles, 140 social-platform posts, 70 company and personal blog posts, 55 from a second academic-paper feed (counted separately from the academic papers), 13 company filings, 5 subscription newsletters and 3 show transcripts. Of those 1,375, 450 were read in full, 364 were screened out as off-topic, 13 were logged with a stated reason for not reading closely, 547 haven't been read yet, and 1 is being refetched.