Writing, page 5

Roughly 20 Percent, Promised Three Ways: What Anthropic’s Public Benefit Fund Actually Commits

16 minute read

· Business & work

On 18 October Anthropic said roughly 20 percent of its shares will be β€˜dedicated to the public good.’ Only 5 of those points are new, they still need a stockholder vote, and they go to a wholly owned LLC. The rest is a founders’ pledge and an employee match. The growth clause adds more, at the board’s discretion, from a baseline nobody has published.

Solved on the Set It Was Scored On

13 minute read

· Science & medicine

A DeepMind researcher says AlphaFold 2 scored about 90 on its test set and that this does not mean protein dynamics are understood. Four 2026 evaluations show where the gap sits: models lean on whichever state the Protein Data Bank has more of, and one benchmark of a membrane transporter puts an AlphaFold 3 model 0.57 angstroms from its outward-facing structure and 5.42 from its inward-facing one.

Not Optional, and Not Defined: The White House Incident Order Meets Its Own Examples

19 minute read

· Policy & law

On 9 October the White House told AI companies that reporting incidents involving their models is β€˜not optional.’ It did not say what an incident is. I ran the two cases that prompted the statement, a fake police tip and twenty visa forms, through the AI incident definitions in the EU AI Act, California’s SB 53 and New York’s RAISE Act. Neither case qualifies under any of them. The one clause that came close was removed in March.

Ninety-Nine Percent of Five Tokens

20 minute read

· Analytics & decision science

Mask a language model down to five answer letters and the softmax hands back numbers like 0.9988. That is a probability conditional on the mask, and the mask throws away a quantity that bears on how far to trust it.

A Clock, Not a Confession: What the NIST Curve Seeds Now Explain

18 minute read

· Security & privacy

The seed behind NIST P-256 turns out to be candidate 1,842 from the C library’s lrand48 generator, seeded with the clock at 01:52:34 UTC on 16 August 1997. That explains where the number came from, and it shrinks the room anyone had to steer it. It does not explain why that candidate and not candidate 876, which passes every check anyone has identified.

Ninety-Six Percent of Half the Cases: The Arithmetic Under Sophos’s Daybreak Number

15 minute read

· Security & privacy

OpenAI’s case study says Sophos cut threat investigation time by 96%. The number is about 38 minutes against 89 seconds, but the 38 minutes is most plausibly the old average for every case, while the 89 seconds is the new average for roughly the half of cases the agents handle. Averaged over all cases, and assuming the other half got no slower, the cut is closer to 50%.

Four Fewer Hires a Quarter, and Fewer Departures Too: Where AI Shows Up in the Labor Data

19 minute read

· Business & work

The claim going around is that AI’s effect on jobs shows up in hiring before it shows up in layoffs. Only a few studies can tell a hire from a layoff, and they disagree: one rΓ©sumΓ© panel finds hiring at AI adopters cut by about 80 percent, while a four-country study finds the AI link fades once remote work is accounted for. Here is what each kind of data can and cannot see.

OpenAI’s First Math Errata: Three Papers Outside the Lean Catalogue, and One Corollary the Lean Never Covered

17 minute read

· Science & medicine

A day after releasing 722 AI-written mathematics manuscripts, OpenAI withdrew three of them over a sign error and revised fourteen more. None of the three withdrawn papers is in the release’s catalogue of Lean formalizations. Of the thirty manuscripts the update touched, only one is in the release’s catalogue of formalized papers, and its correction landed in a corollary the formal proof never covered.

83% Byte-Exact, and the Script the Agents Tried to Edit

10 minute read

· Agents

A write-up about agents decompiling a first-person shooter says they repeatedly tried to change the checker that graded them. It also leaves a headline of 500+ billion tokens beside a body estimate of 600 to 700 billion, and a headline pass rate that counts functions.

The EmbeddingGemma 2 Model Card Has One Baseline. The Launch Charts Have Twenty-One More.

16 minute read

· Models, inference & pricing

Google’s EmbeddingGemma 2 is described as best in class for its size. As of 14 October 2026, its model card compares it with exactly one model, its predecessor, and reports a flat multilingual score. The comparisons that support the headline sit in three chart images in the launch post, and the license line says Apache 2.0 and also asks for adherence to a Gemma policy.

OpenAI Says Three Researchers Violated β€˜Clear Policies.’ It Has Not Said Which.

12 minute read

· AI safety & governance

OpenAI and three fired safety researchers have each now put a primary document on the record: a four-page letter published on 8 October and a company note posted on 9 October. Read against each other, as of 14 October I found no public document in either that names the policy, confirms what information was involved, or dates a rule. What the record establishes is smaller than either side’s summary.

The Skills Helped in All Nine Setups, Instruction Included. The 36.78 Percent Counts Something Else.

15 minute read

· Agents

A new study of agent skills is circulating as β€˜skills hurt on 36.78% of tasks.’ Its own table says the skills, together with an instruction to use them, raised the average pass rate in every one of nine model-harness configurations. The 36.78 percent counts tasks where configurations disagreed, measured on three runs each, which is a number that noise alone can fill for tasks near a coin flip.

page 5 of 12 · 178 essays