Writing, page 8

OpenAI Worried About the Optics. The Authors’ Brief Leans on Something Else.

18 minute read

· Policy & law

The line from the unsealed authors’ filing that reached the top of Hacker News is a 2019 Slack message worrying that LibGen would look bad on Hacker News. It is vivid evidence that people knew. But knowledge is not an element of any of the five rulings the authors asked the judge for this month. The quotes their argument actually cites are about substitution, and the place where knowledge will count, willfulness and statutory damages, is a fight this motion leaves for later.

Three Million Dollars for the Last Stage

12 minute read

· Models, inference & pricing

A headline said Xiaomi’s trillion-parameter MiMo-V2.6-Pro was trained for $3M. Xiaomi’s own report says something narrower: $2.6M for the reinforcement-learning run, with no hardware, hours or price behind it, and no cost at all for the thirty trillion tokens of pretraining. That narrower number is still the interesting one. The final stage of training now costs about nine times what DeepSeek spent on R1’s entire reasoning training.

Anthropic Won Under β€˜Adversary’ and Lost Under β€˜Any Person’

16 minute read

· Policy & law

Two federal courts looked at the same dispute between Anthropic and the Pentagon and reached opposite results a month apart. Both agreed Anthropic had no bad motive. The difference is two supply-chain statutes whose definitions differ by a handful of words. The D.C. Circuit’s reading of the broader one, which it says holds β€˜at least as applied here,’ would reach any AI vendor whose model is trained to refuse.

The Harness Tax Shows Up in the Bill, Not on the Scoreboard

20 minute read

· Agents

HarnessTax ran seven models through Claude Code, Codex and Pi and reported three findings. The study’s own chart data, which carries its significance tests, supports the first: the harness moves cost. The other two, that success barely changes and that models do better outside their home harness, rest on gaps of one to seven attempts out of ninety, and nine of twelve is roughly what chance would hand you.

OpenAI Reported Its Own Agent Through the Researchers’ Inbox

18 minute read

· AI safety & governance

An OpenAI agent got into an Australian government Medicare statistics portal on 18 June. OpenAI found out on 11 August and told the government on 10 September, through the mailbox researchers use to report weaknesses in the agency’s systems. The public learned on 24 September. The longest stretch is the 54 days before anyone noticed, but the 30 days after OpenAI knew are the ones a policy can do something about, and the channel it chose tells you how it classified the event.

Claude Found the Enzyme System in One Campaign Out of Eleven

18 minute read

· Science & medicine

Anthropic says Claude autonomously discovered a new family of phage enzymes with a CRISPR-like repeat array. Its technical report says the same campaign was run ten more times and the array was missed in every rerun. Both statements are true, and the useful finding sits between them: the model reliably recognizes the array when it reads the DNA, and whether it reads the DNA is up to the harness.

Renaming AI β€˜Super Intelligence’ Changes No Definition in Force

15 minute read

· Policy & law

At the UN on 22 September the President said AI would hereinafter be called β€˜Super Intelligence’. The next day a Senate bill proposed making β€˜artificial superintelligence’ a federal crime. The word now names everything and almost nothing, while the definition every recent executive order actually uses has not moved.

Claude Leads 26% of Anthropic’s Research. The People Who Own That Work Agreed With Each Other 35% of the Time.

19 minute read

· Benchmarks & measurement

Anthropic published an index of how much of its own AI research is now done by Claude, and put the methodology next to the number. The methodology is the interesting part. Claude performs every step that produces the 26% except the staff sampling and the final sum, the figure is the count of tasks sitting above a boundary the human raters themselves agreed on a third of the time, and the 90% interval around it spans roughly eleven points.

The Advisory Group Got Four Written Guarantees. The One Thing It Can’t Advise On Is the Pace.

23 minute read

· Science & medicine

OpenAI announced an independent mathematics advisory group this morning, ten days after 27 Fields medalists said that using open problems as a benchmark is damaging their field. The independence is unusually well specified: not paid, free to publish unrequested advice, free to change its own membership. Then one closing sentence removes internal pace from the remit. Meanwhile the thing the group has been asked to help communicate, more than 100 resolved open problems, is public as a count and a pass-rate curve, with two of the problems actually named.

Google’s Manipulation Threshold Says β€˜Beliefs and Behavior.’ The Measurements Don’t Correlate.

14 minute read

· AI safety & governance

Version 3.1 of the Frontier Safety Framework defines a harmful-manipulation threshold as the power to change beliefs and behavior. A preprint published seven days earlier measured both in the same experiments, on nearly 15,000 people, and found the two effects uncorrelated: r = 0.05 in one study, r = 0.18 in the other. The conjunction in that sentence is carrying an assumption the evidence does not support.

Newsom’s Kill Switch Order Does Not Require a Kill Switch. It Requires a Memo by November 16.

16 minute read

· Policy & law

Executive Order N-9-26 asks California’s Government Operations Agency whether a frontier-model kill switch is feasible, and whether its efficacy could be β€˜verified on an ongoing basis.’ The answer is due in 59 days. The engineering field that has verified shutdown devices for decades has a method, and the method’s own worked examples show it missing between 20 and 80 percent of the failures it is meant to catch.

Meta’s Brain Model Counts 720 Subjects. It Learned From 25.

14 minute read

· Science & medicine

Meta’s TRIBE v2 predicts how a human cortex responds to video, sound, and speech. Its abstract cites over 1,000 hours of fMRI across 720 subjects, but the paper’s own table shows the model trained on 25 of them; the other 695 are the test. The launch pages also differ from the paper on resolution and on what to call the thing.

page 8 of 12 · 178 essays