R[AI]DAR

Category intelligence

AI News Briefing

All 19 current items, analyzed and ranked.

Executive synthesis

AI News Summary

Executive Signal

  • Political rebranding of AI as "Super Intelligence" and rapid frontier-model proliferation are racing ahead, while systemic risks—security signal decay and workforce inequity—compound beneath the surface.

Priority Developments

  • Trump's "Super Intelligence Force" couples a centralized AI safety task force with a non-binding industry pact, signaling a pivot from technical governance to narrative management.
  • Frontier models from OpenAI, Anthropic, and Google shipped within 30 days; benchmarks converge but pricing, access tiers, and per-task cost diverge sharply, reshaping AI procurement calculus.
  • NASA and IBM's open-source Lunar Foundation Model demonstrates domain-specific foundation models can outperform generalists on scientific prediction, opening a new play for vertical AI.
  • Google's pause of its open-source bug bounty reveals generative AI is flooding security workflows with low-quality submissions, eroding trust signals across the research ecosystem.
  • Women hold a small share of new AI roles while over-indexing in displacement-exposed jobs, signaling the AI economy is widening structural inequality, not closing it.

Leadership Implications

Key Themes

US AI policy under Trump · 4AI safety and governance · 3Open source releases · 4Frontier model landscape · 2AI research advances · 2Geopolitical AI dynamics · 1AI infrastructure and community impact · 2

Primary evidence

All Ranked Signals

Showing 10 of 19 signals 2 remaining
News AI News & Artificial Intelligence | TechCrunch 2 days ago

Trump unveils his new Super Intelligence Force

By Anthony Ha

70 score
AI Analysis

Continuing our coverage from yesterday, Trump has launched a 'Super Intelligence Force' task force on AI safety, led by DNI Jay Clayton and reporting directly to the president. The unit coordinates with AI companies, critical infrastructure operators, and interest groups.

Source details
This new task force is Trump's latest response to the debate over AI safety.
US AI policyAI safetyexecutive actionnational security
News The Decoder 3 days ago

NASA and IBM's open source lunar model turns 17 years of orbiter data into a foundation for lunar science

By Jonathan Kemper

62 score
AI Analysis

NASA and IBM have released the Lunar Foundation Model, an open-source AI model trained on roughly 17 years of Lunar Reconnaissance Orbiter data that reduces error in predicting polar ice deposits. It is positioned as a foundation model for lunar science.

Source details
NASA and IBM have released the Lunar Foundation Model, one of the first open-source AI models for lunar science. Trained on nearly 2 million tile bundles, mostly from 17 years of Lunar Reconnaissance Orbiter data, it cuts the error in predicting polar ice deposits. The article NASA and IBM's open source lunar model turns 17 years of orbiter data into a foundation for lunar science appeared first on The Decoder.
open sourcescientific AIfoundation modelsNASAIBM
News The Decoder 2 days ago

Trump launches "Super Intelligence Force" that has nothing to do with actual superintelligence

By Matthias Bastian

60 score
AI Analysis

Continuing our coverage from yesterday, The Decoder notes that Trump's 'Super Intelligence Force' uses 'superintelligence' as a synonym for AI rather than as a technical term, and reports the unit is led by DNI Jay Clayton with direct presidential reporting lines.

Source details
Donald Trump has established a "Super Intelligence Force." "Superintelligence" is his term for artificial intelligence. The unit is led by Director of National Intelligence Jay Clayton and other government officials, coordinates with AI companies, critical infrastructure operators, and interest groups. The group reports directly to the president. The article Trump launches "Super Intelligence Force" that has nothing to do with actual superintelligence appeared first on The Dec
US AI policyexecutive actionAI safety
News MarkTechPost 2 days ago

GPT-6 Astra vs GPT-6.1 Sol vs Gemini 4 Argon vs Claude Fable 5.1: Which Frontier Model Fits Which Job

By Asif Razzaq

60 score
AI Analysis

MarkTechPost compares four frontier models that shipped within a 30-day window: Claude Fable 5.1, GPT-6 Astra, GPT-6.1 Sol, and Gemini 4 Argon, noting OpenAI canceled GPT-6.1 Astra on September 28. Pricing, access rules, and cost per task diverge sharply even as benchmark scores converge.

Source details
Anthropic, OpenAI and Google DeepMind shipped 4 frontier-class models within 30 days. Claude Fable 5.1 arrived on September 1. GPT-6 Astra followed on September 3. GPT-6.1 Sol and Gemini 4 Argon landed in the last days of September. We covered each launch on its own. This piece puts them side by side. The benchmark scores overlap more than the launch posts suggest. The prices, access rules and cost per task do not. One change frames the lineup. OpenAI cancelled GPT-6.1 Astra on September 2
frontier modelsmodel comparisonpricingOpenAIAnthropicGoogle
News AI (artificial intelligence) | The Guardian 2 days ago Old anchor

Trump names intelligence chief Jay Clayton as new White House AI czar

By Ariana Baio

55 score
AI Analysis

Continuing our coverage from yesterday, President Trump has appointed Director of National Intelligence Jay Clayton as White House AI czar, consolidating AI policy leadership under a sitting intelligence chief. The move signals that the administration is folding AI coordination into the national security apparatus.

Source details
Clayton had said AI was a ‘gamechanger’ but also posed ‘a threat’ during his confirmation hearingDonald Trump on Sunday named Jay Clayton, the director of national intelligence, to also serve as the new White House AI czar.Clayton, who leads the US’s 18 intelligence agencies, previously served as the top federal prosecutor in Manhattan and led the Securities and Exchange Commission during Trump’s first term. Continue reading...
US AI policynational securitygovernment appointments
News MarkTechPost 3 days ago Old anchor

Aleph Alpha Releases Kolibri: A 78.1B Open-Weight English-German MoE Model With Only 3.46B Active Parameters

By Asif Razzaq

55 score
AI Analysis

Aleph Alpha has released Kolibri, a bilingual English-German MoE model with 78.1B total parameters, only 3.46B active per token, 1M-token context, and Apache 2.0 licensing. An FP8 checkpoint fits a single B200/B300/H200 or two H100 SXM5 GPUs via vLLM.

Source details
Aleph Alpha has released Kolibri, an open-weight Mixture-of-Experts (MoE) language model built for German and English. Kolibri has 78.1B total parameters but activates only 3.46B, or 4.4%, per token. It accepts up to 1,048,576 tokens of context, lets users set reasoning effort per request, and ships under the Apache 2.0 license on Hugging Face. The target is sovereign deployment in regulated sectors such as public administration, industry and aerospace. Is it deployable? Yes. The FP8 checkpoi
open sourceopen weightsMoEAleph Alphasovereign AIEuropean AI
News MarkTechPost 3 days ago Old anchor

Google Research Moves Federated Learning Into TEEs: Gboard Now Trains With Externally Verifiable Differential Privacy

By Michal Sutter

55 score
AI Analysis

Google Research has built a next-generation Federated Learning system on Trusted Execution Environments, claiming externally verifiable central differential privacy guarantees for the first time. It underpins Gboard next-word prediction and related on-device features.

Source details
Google Research has announced a next-generation Federated Learning (FL) system built on Trusted Execution Environments (TEEs). The research team claims externally verifiable central differential privacy (DP) guarantees for FL for the first time. What Problem Does TEE-Based Federated Learning Solve? Google introduced Federated Learning FL in 2017. It powers next-word prediction and Smart Compose on Gboard, reply suggestions in Google Messages, and Smart Text Selection in Android. Earlie
federated learningprivacydifferential privacyGoogle researchon-device AI
News The Decoder 2 days ago Old anchor

Google researchers find a way to keep self-improving AI agents from memorizing their tests

By Jonathan Kemper

55 score
AI Analysis

Google researchers propose RRSI, a method that limits how much self-improving AI agents memorize their training tests, lifting scores on unseen benchmarks by up to 4.7 points while using about 30% fewer tokens than an unregularized baseline.

Source details
Self-improving AI agents tend to memorize their test tasks, so their gains shrink or disappear on new ones. RRSI, a new method from Google researchers, reins in this effect and lifts scores on unseen benchmarks by up to 4.7 points while using about 30 percent fewer tokens than an unregularized version. The article Google researchers find a way to keep self-improving AI agents from memorizing their tests appeared first on The Decoder.
AI researchself-improving agentsevaluationGoogle
News The Decoder 3 days ago Old anchor

Chinese AI models parrot state doctrine or refuse to answer on sensitive topics

By Manuel Uth

55 score
AI Analysis

An Aleph Alpha study finds leading Chinese AI models echo state doctrine or refuse to answer on politically sensitive topics, with only 17-41% of responses rated balanced. Aleph Alpha, which sells 'sovereign AI' to governments, has a commercial interest in highlighting the contrast.

Source details
Chinese AI models often follow the party line on politically sensitive questions, according to an Aleph Alpha study that rated only 17 to 41 percent of answers as balanced. Aleph Alpha sells "sovereign AI" to governments, giving it a commercial interest in distinguishing its models from Chinese competitors. The article Chinese AI models parrot state doctrine or refuse to answer on sensitive topics appeared first on The Decoder.
Chinese AIcensorshipsovereign AIAleph Alphageopolitics
News The Decoder 3 days ago

Google's new Gemini tiers cut free users to its weakest model and lock $5/month subscribers out of Pro

By Matthias Bastian

52 score
AI Analysis

Starting in October 2026, Google restricts Gemini access: free users will only get Flash-Lite, while Flash and Pro are reserved for paying tiers. The change is read as preparation for the more resource-intensive Gemini 4 Argon.

Source details
Starting in October 2026, Google will cut free access to its Gemini models. Users without a subscription will only get the smallest model, Flash-Lite, while Flash and Pro will be reserved for paying customers. The move could also set the stage for the more resource-hungry Gemini 4 Argon. The article Google's new Gemini tiers cut free users to its weakest model and lock $5/month subscribers out of Pro appeared first on The Decoder.
GoogleGeminiproduct pricingaccess tiers