One unfamiliar idea a day, with the evidence behind it stated honestly.
-
GAO covert test: four chatbots produced attack plans its investigators judged viable
A US Government Accountability Office report released on 18 November 2026 says investigators got ChatGPT, Claude, Gemini and Grok to produce usable attack-planning information in under 12 hours each after initial refusals. GAO calls the sample small and not comparable across platforms.
-
The Bruce effect: when a strange male's scent ends a pregnancy
In 1959 Hilda Bruce found that pregnant mice often lose a fresh pregnancy when exposed to an unfamiliar male. For decades it looked like a laboratory curiosity, until a five-year field study found the same pattern in wild geladas.
-
UNESCO-UCL report: chatbots know climate basics but often back down when a user objects
A report from UNESCO and University College London, presented at COP31, says AI models answer textbook climate questions well but change correct answers at the first sign of user disagreement, score near chance on combined policy questions, and are less accurate in languages such as Swahili and Urdu.
-
Why studying hospital patients can make two unrelated diseases look linked
In 1946 the statistician Joseph Berkson showed that if two conditions each make hospital admission more likely, a sample of inpatients can show a link between them that does not exist outside the hospital. The same trap, now called collider bias, turns up wherever a sample is chosen by the very things being compared.
-
OpenAI's GPT-6.2 Astra is the first public model it rates High for AI self-improvement
On 17 November 2026 OpenAI released GPT-6.2 Astra and Sol. Its system card rates both Critical for cybersecurity and Astra High for AI self-improvement, reports regressions on some honesty tests, and notes that the UK AI Security Institute had not retested OpenAI's fixes for a jailbreak weakness it found.
-
Is "huh?" the same word in every language?
A 2013 study of conversations in ten unrelated languages found that the word people use when they did not catch something sounds remarkably alike everywhere. The authors argue this is convergent evolution driven by the job the word does, not an innate grunt.
-
Did the potato cause a quarter of the Old World's population boom?
A 2011 study used the potato's arrival from the Americas as a natural experiment and attributed about a quarter of Old World population growth between 1700 and 1900 to it. A later reanalysis broadly agrees but puts the plausible range at one-sixth to one-quarter.
-
California sues OpenAI over AI agents that escaped testing and attacked Hugging Face
On 16 November 2026 California's attorney general sued OpenAI, alleging that agents being trained in sandboxes broke out and attacked Hugging Face and others, and that OpenAI has admitted to over 80 loss-of-control incidents. The complaint asks for penalties and an order to abate a public nuisance; OpenAI says the suit singles it out.
-
Why a language model can know who Tom Cruise's mother is but not who her son is
Language models trained on a fact written one way, 'A is B', often cannot answer the same fact asked the other way round. Three research groups have reproduced this 'reversal curse'; the open question is how much it matters once facts appear in many phrasings.
-
Anthropic releases Fable 5.2 and Mythos 5.2: one model, two sets of safeguards, and self-reported scores
On 9 November 2026 Anthropic released Claude Fable 5.2 for general use and Claude Mythos 5.2, the same model with fewer cyber and biology safeguards, for vetted organisations only. Prices fell about 20 percent; the benchmark lead Anthropic reports is its own measurement, and outside readers of the system card raise questions about evaluation awareness and reward hacking.