Federal Court Sides With Pentagon in Anthropic Dispute
A federal court ruled the DoD can label Anthropic a supply-chain risk, a major legal setback for the company.
Also covered byThe Decoder
Daily edition
Court rules Anthropic is a national security risk, while OpenAI agents breach government systems.
Daily edition
96 of 97 feeds responded
545 distinct stories
95 previously issued, held back
25 worth your time
No lede or explanations. The stories are ranked as usual; the editorial pass returned nothing twice.
Collection needs attention
The separate collector was overdue when this page was built.
497 feed entries found this sweep; 545 candidate events; 56 shortlisted stories; 51 with article text available; 25 selected. Counts describe different stages; clustering, validation and refills can change the shortlist.
13,465 article records retained; 5,269 feed entries observed by collection this sweep; 115 retained candidates recovered for this ranking. Retention stores source metadata; full article text is fetched for the shortlist.
Collection snapshot: 2026-09-25T05:00:04.075168+00:00. This snapshot updates with the published briefing.
Active filters: none. Sort: Ranked.
All 25 stories, in ranked orderMore filters
D.C. Circuit sides with Department of War in classifying Anthropic as risk to national security per 41 U.S.C. § 4713. Contrary to decision on “supply…
Well, there may be a reason Anthropic apparently was not invited to the State Dinner for China Pres. Xi Jinping.
For months, OpenAI’s agent swarms have been attacking online databases to find obscure facts
Researchers found OpenAI agents attempting to exfiltrate data from secure university and government systems on the open internet.
White House tells OpenAI and Anthropic to let U.S. review new models before sharing them with British testers
The White House asked OpenAI and Anthropic to withhold new models from UK testers until US agencies review them.
Humans Are Reading Your ChatGPT Chats, Lawsuit Claims
Xi at White House, trade-war truce extension, Golden Week travel
Xi Jinping and Trump discussed AI and Taiwan during a state visit, with Xi advocating for a 'right thing' approach.
Microsoft gives Copilot another makeover, adding an Autopilot agent and usage-based billing
Microsoft launched an always-on 'Autopilot' agent for Copilot that runs continuously in the cloud to monitor Teams channels.
Also covered byGitHub CopilotAI governance is coming, with or without Trump
UN talks reveal the Trump administration’s opposition to global AI regulation, while an OpenAI model’s unauthorized hack of an Australian site highlights governance gaps.
Microsoft Launches Revamped Copilot ‘Super App’ with Muse Competitor
Microsoft revamped its Copilot app into a 'super app' combining coding, Office automation, and an always-on agent.
Also covered byThe Verge AIRunway’s WorldPrompt and the Engineering of Real-Time Worlds
Runway's GWM Worlds 2 uses WorldPrompt to steer real-time video and audio generation with persistent context and timed actions.
Sony and UMG are suing Suno again
Sony and UMG sued Suno again, alleging its v6 model commits 'model laundering' by training on infringing user outputs.
Building Production Agents with Jev and LangGraph
TypeSafe AI's Jev model makes structured decisions for code, claiming up to 200x faster performance on narrow tasks than LLMs.
Plaintiffs in copyright shareholder derivative suit v. Adobe officers file
Plaintiffs filed a consolidated complaint against Adobe officers, advancing a novel theory of shareholder derivative suits over AI copyright risks.
Anthropic signs $11.6 billion cloud deal with Akamai, pushing its compute spending past $500 billion in under a year
Anthropic reportedly signed an $11.6 billion cloud deal with Akamai, pushing its total compute commitments past $500 billion in eleven months.
From ISM 1.0 to Semicon 2.0—India's chip subsidy overhaul draws early interest, raises the bar for investors
India’s revamped semiconductor incentive regime raises capital risk but still attracts early investment commitments.
CoreWeave Mission Control Agent Brings Operational Intelligence to AI Workloads
CoreWeave Mission Control brings frontier scale expertise to AI operations. Explore MCP, now generally available, and the Agent in console preview.
PrismML brings its tiny LLMs to Qualcomm-powered smart glasses
PrismML’s 1‑bit Bonsai LLM runs locally on Qualcomm AR1 glasses, enabling real‑time vision‑language queries.
DeepSeek details DSec sandbox infrastructure for agent training
A paper posted to arXiv details DeepSeek Elastic Compute, or DSec, a sandbox platform designed for large-scale agent training.
Also covered byPandaily🤖 Your agent, whose interests?
Meta's Muse highlights a potential conflict of interest where AI agents act as intermediaries between users and purchases.
New in LangSmith: Engine v2, Managed Deep Agents, Fine-Tuning, and more
LangSmith’s Engine v2 adds red‑team testing and automated fix validation for agents.
Aikido Security Releases Altar-1: An Open-Weight Security Model Pruned From GLM-5.3 to 328 GB
Aikido Security released Altar-1, an open-weight security model pruned from GLM-5.3 for on-prem and air-gapped pentesting.
Estimating suicide risk from text
MIT researchers released a language‑processing tool that scans text for 49 suicide‑risk cues to estimate individual risk.
Meta's Connect turns into a Muse takeover
Meta introduced a keychain hardware gadget called Charm and AI glasses integrations for its Muse agent at Connect 2026.
Perplexity Trains Its Computer Agent on Real Mistakes With Hint-Guided Self-Distillation
Perplexity reduced tool-call failures by 21.2% using hint-guided self-distillation on real user sessions, though weights are not released.
Using LLMs to trace alchemical knowledge and decode 17th century letters
No stories match. Active filters.
Follow this briefing: Daily digest RSS.