AI news, 5 October: Trump launches Super Intelligence Force to steer U.S. AI policy

Trump forms Super Intelligence Force. President Donald Trump announced on Sunday the creation of a “Super Intelligence Force” to coordinate the federal government’s work on artificial intelligence, which his administration has rebranded as “super intelligence.” According to TechCrunch and The Washington Post, it will be chaired by Director of National Intelligence Jay Clayton, with Federal Trade Commission Chair Andrew Ferguson, Pentagon technology chief Emil Michael and Office of Personnel Management Director Scott Kupor as vice chairs. The group reports to Trump and has roughly 120 days to produce a report on risks and opportunities, with a charter emphasizing U.S. leadership while avoiding overregulation. The move follows a recent voluntary safety accord signed by major tech companies at the White House.[[1]](https://techcrunch.com/2026/10/04/trump-unveils-his-new-super-intelligence-force/)

Google restricts free Gemini models. From 9 October, free users of Google’s Gemini app will be limited to the lighter Flash-Lite model and lose access to the more capable Flash and Pro versions, according to an updated Google support document reported by TechRadar and others. Subscribers to the roughly $5-a-month AI Plus plan will keep Flash but lose Pro. Higher-paid tiers are unchanged and gain a “Deep Think” reasoning mode. The shift brings Gemini’s free tier closer to the freemium approach used by ChatGPT.[[2]](https://www.techradar.com/ai-platforms-assistants/gemini/if-youre-on-the-free-or-plus-tiers-for-google-ai-youre-about-to-lose-access-to-some-gemini-models-heres-whats-changing)

Aleph Alpha opens Kolibri model weights. German AI firm Aleph Alpha has released Kolibri, a 78-billion-parameter mixture-of-experts model focused on German and English, with full weights available under the Apache 2.0 license. It supports context windows up to one million tokens, controllable reasoning and tool use, while activating only about 3.5 billion parameters per token for efficiency. The company positions it for sovereign, mission-critical applications in government, industry and aerospace.[[3]](https://www.marktechpost.com/2026/10/04/aleph-alpha-releases-kolibri-a-78-1b-open-weight-english-german-moe-model-with-only-3-46b-active-parameters/)

Amazon commits $1 billion to data-center communities. Amazon said it will add more than $1 billion over five years through its “Built Together” program for U.S. towns and counties that host its data centers, funding community college access, job training, home and school energy upgrades, and locally chosen projects. AWS CEO Matt Garman framed the pledge as a response to growing local opposition and more than 100 proposed moratoriums that he warned could undermine America’s AI lead, according to Reuters and company statements.[[4]](https://www.reuters.com/business/retail-consumer/amazon-invest-1-billion-over-five-years-us-data-center-communities-2026-10-02/)

NASA and IBM release open lunar science model. NASA and IBM have made available an open-source foundation model trained largely on 17 years of data from the Lunar Reconnaissance Orbiter plus other missions. Hosted on Hugging Face, it helps researchers map craters, study volcanic features and locate potential polar ice deposits, cutting error rates on ice prospecting by up to 22 percent versus strong baselines and supporting future crewed Moon exploration.[[5]](https://the-decoder.com/nasa-and-ibms-open-source-lunar-model-turns-17-years-of-orbiter-data-into-a-foundation-for-lunar-science/)