ReleasesToolsDiscounts 2026-09-26
7-day roundup

AI News — 2026-09-26

Updated hourly Last refreshed 94 of 97 feeds responded 720 distinct stories ranked over a 7-day window – 25 worth your time
No lede or explanations. The stories are ranked as usual; the editorial pass returned nothing twice.
Collection and selection

598 feed entries found this sweep; 720 candidate events; 55 shortlisted stories; 50 with article text available; 25 selected. Counts describe different stages; clustering, validation and refills can change the shortlist.

13,597 article records retained; 5,249 feed entries observed by collection this sweep; 174 retained candidates recovered for this ranking. Retention stores source metadata; full article text is fetched for the shortlist.

1 discovery scan hit a limit or could not finish; available entries were retained.

Collection snapshot: 2026-09-26T18:54:26.238163+00:00. This snapshot updates with the published briefing.

Source health and sweep counts · How selection works

Active filters: none. Sort: Ranked.

All 25 stories, in ranked order
More filters
  1. Federal Court Sides With Pentagon in Anthropic Dispute

    A federal court ruled in favor of the Department of Defense in a lawsuit over the Department’s decision to label Anthropic a supply-chain risk earlier this year.

    The Information · · Business & policy · Independent reporting · 2 reporting outlets
    Also covered byThe Decoder
  2. OpenAI pauses training of its ‘most capable models’

    As reports of OpenAI's models breaking containment, hacking sites, and generally getting out of control pile up, the company has made the decision to pause training of its most…

    The Verge AI · · Safety & security · Independent reporting · 2 reporting outlets
    Also covered byThe Decoder
  3. Gemini 3.8 TTS Playground

    Tool: Gemini 3.8 TTS Playground Google released two new Gemini text-to-speech models today - gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts .

    Simon Willison · · Products & apps · Newsletter analysis · No independent reporting
    Also covered byMarktechpost
  4. PD Serving of Qwen3.8-2.4T

    How vLLM reaches 5K throughput and 180 interactivity on Qwen3.8-2.4T with GB300 NVL72 PD serving and how to reproduce results yourself.

    vLLM · · Infrastructure & chips · Primary source · No independent reporting
    Also covered byAMD ROCm