Newspapers, programmes, reviews and conversations. Who is speaking, where they appear, and what is being debated.
Publication records for NHK, TBS, TV Asahi and BBC Japanese.
Explore records →Hinton’s warnings, AI 2027, and disagreement about accelerating development.
2025 records →A Japanese book edition, author interviews, newspaper reviews and public discussions.
Explore records →This sequence organises the collected sources. It does not establish influence between outlets or measure public awareness.
Reporting on the statement signed by over 350 researchers and executives, also covering criticism from Yann LeCun and others.
The public page and identifying metadata were checked. This does not mean its entire body or claims have been validated.
Checked: 2026-09-08
Same event / related publications: R14 · R15 · R16 · R36 · R56
All 49 entries in the supplied bibliography are included, plus seven additional records from source searches. This is not an exhaustive search of print, paywalled articles or broadcast archives. The 56 records are not 56 independent news reports.
Newspapers, broadcaster web output, translated journalism, publisher promotion, individual commentary and events are distinguished. Publication volume cannot establish how many people understand the arguments.
English accompanies the Japanese descriptions. English titles are editorial renderings, not official English headlines from the outlets. The images page pairs two Asahi headline excerpts and the top of a TBS article with English versions. Full article translations are not included.
Japanese excerpts were captured after checking their sources in a browser. English images are unofficial editorial translations, not English editions issued by the outlets. Hidden text has not been reconstructed.
Soares and Yudkowsky’s strong extinction predictions, Kokotajlo’s alternative scenarios and Hinton’s risk warnings are distinct claims. Extinction, catastrophe and loss of human control are not interchangeable.
Figures such as “88% accurate” and “70%” depend on their sources’ definitions and assessments. They are not treated here as validated statistics. Agreement with an argument is different from understanding it.