Read online version of the 25 September 2026 issue
Newsletter/25 September 2026
The labs asked the Security Council to take the decisions off them
Altman and Amodei at the UN on Wednesday; what Claude's 'CRISPR-like' enzyme find actually found; and the re-price test for Tuesday's price cuts.
By Daily Aletheia · Checked against the primary source · 25 September 2026 · 4 min read

On Wednesday the two men who sell the most capable AI models asked the UN Security Council to take the biggest decisions about them away from the labs, and one of those models had just been credited with a discovery whose function nobody knows.
The Signal - The labs asked the Security Council to take the decisions off them
On Wednesday 23 September Sam Altman addressed the UN Security Council, and OpenAI published his remarks as delivered. He asked for national and international standards for measuring AI capabilities, assessing risks and judging whether safeguards are sufficient; for incident-reporting protocols "so the world can learn from failures before they become catastrophes"; and for secure channels between governments, infrastructure operators and technical experts. "If AI is to be democratic," he said, "the most important decisions cannot be made by labs in San Francisco alone." He also said OpenAI "should not train models that we cannot make an extremely strong case that we will be able to keep under human control," and that it has "unilaterally slowed down in the past" and will do so again. Anthropic's Dario Amodei addressed the Council the same day; CNN reports he called AI "the most important global security issue facing the world today" and asked members to start with narrow agreements, such as a ban on using AI to make biological weapons. Anthropic has not published his remarks.
The standards Altman listed - capability measurement, incident reporting, human oversight - are the clauses your AI contracts inherit next. A vendor that asks for them in public can be held to them in your procurement questionnaire, and you can start asking now. And a seller who says on the record that it will slow down when it must is telling you how to buy: on what ships today, never on a roadmap.
The counterweight: the day before, on Tuesday, the same two companies shipped new frontier models and cut their prices. Anthropic's Opus 5.5 page calls it "our first release since we called for pacing the frontier." The call to pace and the launch came in the same month.
Free to keep reading. Every issue, checked, three mornings a week.
Three mornings a week, free.
The Move - the re-price test
The Move - the re-price test is for members. $9 a month, founding price.
The check
- What was said
- "Claude found a CRISPR-like enzyme system that experts missed."
- What the source says
- Claude agents (about 950, 21 hours, 210M tokens) noticed a repeat array and a partner protein beside a reverse transcriptase already identified in earlier studies; Anthropic's scientists confirmed the array is expressed as short RNAs; the function is unknown; no editing shown; all lab work by humans; pre-print only.
- The difference
- The pattern is real and newly noticed; the enzyme was known, 'CRISPR-like' is the layout not the function, and what it does is unknown. Verdict: half true.
Sources & further reading
- 01OpenAI - Sam Altman's remarks at the United Nations Security Council, as delivered (23 Sept 2026) ↗Primary
- 02Anthropic - Claude discovers a novel enzyme system with CRISPR-like repeats (23 Sept 2026) ↗Primary
- 03Anthropic - Introducing Claude Opus 5.5, pricing table (22 Sept 2026) ↗Primary
- 04OpenAI - Introducing GPT-6 Sol and Luna (22 Sept 2026) ↗Primary
- 05CNN - Sam Altman, Dario Amodei urge UN Security Council to adopt international AI standards (23 Sept 2026) - Amodei's remarks ↗Attributed
Was this issue useful?