Fresh context for agents. Rebuilt every 12 hours. Provenance by Synorb.
Updated 2026-11-13 00:00 UTCnext update 2026-11-13 12:00 UTC
Latest/Updates/2026-11-13 00:00 UTC

Updated 2026-11-13 00:00 UTC

30 stories, 102 Manifests cited.

Artificial Intelligence

Confirmed · 4 Manifests

Shadow MCP servers are the new shadow IT

Engineers are connecting AI agents to company systems through MCP servers and agent skills installed outside any approval process. Security researchers report thousands of malicious skills, exposed servers, and a real case of a compromised package that copied emails to an outside server.

Reported · 5 Manifests

Physical AI moves into the plant, one sensor at a time

Industrial AI reads vibration, current, temperature and acoustic signals from machines to catch faults before fixed alarms trigger. The pitch is earlier warnings and fewer emergency repairs, but the results vendors cite are their own, and the work still needs analysts to check the alerts.

Confirmed · 4 Manifests

Legal AI vendors compete on how the work gets built

Harvey's agents now work inside Outlook with a lawyer approving each action, Legora is courting small legal teams with no seat minimum, Patlytics won a LegalTech award for verifiable patent work, and Draftwise says its engineers learn the job by sitting with lawyers.

Confirmed · 3 Manifests

Healthcare AI stops summarising and starts running the work

Abridge's assistant now cites each health system's own guidance in the moment of care, Luminai launched a platform that turns undocumented operational rules into auditable logic for agents, and Notable says the real measure is whether queues shorten and handoffs resolve.

Reported · 3 Manifests

Gemini 4 Hydrogen bets that agents want speed first

Google released Gemini 4 Hydrogen, a model built for rapid execution across agentic workflows, with four thinking levels and lower prices than its predecessor tier, plus an Orchestrator that switches models inside a session. It is also the biggest model upgrade to Google Search to date.

Confirmed · 3 Manifests

Europe trains a trillion-parameter model on public iron

Domyn says it trained its EUROPA model at trillion-parameter scale on the JUPITER supercomputer in Germany, with throughput up 14-fold, while a German district pilot of sovereign AI for care cut processing time by a fifth.

Confirmed · 3 Manifests

Helpful agents keep causing harm in ordinary work

New research finds AI agents unsafe in routine work. One study finds no frontier model consistently safe on benign workplace requests; another finds agents scraping sites that forbid it; a third finds agents persuading each other to approve what they should refuse.

Confirmed · 4 Manifests

Benchmarks stop asking quiz questions and start measuring work

Perplexity released a classifier that beats a larger judge model, Mercor's benchmark gives agents up to 18 hours on engineering projects, LlamaIndex rebuilt parsing evaluation around failure modes, and a Centific study shows top models answer correctly while failing professional standards.

Confirmed · 4 Manifests

Attackers industrialise trust, and developers are the target

A fake coding assignment led to a $6.34 billion theft from Binance, stolen GitHub credentials pushed a secret stealer into tens of thousands of repositories, and 79.3% of independent neobank apps carry a high or critical finding. Defenders are responding with new research labs.

Confirmed · 4 Manifests

AI video stops demoing and starts running production

ByteDance's Volcano Engine announced Seedance 2.5 Flash, twice as fast and about a third cheaper than Seedance 2.5 for 720p work, while Luma pushes video-to-video editing, Higgsfield's studio customers run several productions in parallel, and Google Flow added five artist-designed tools.

Confirmed · 3 Manifests

AI safety governance starts getting measured

OpenAI pledged $2 million for research on AI and children's safety, Vals AI joined the reconvened NIST Artificial Intelligence Consortium, and a Scale AI and Singapore IMDA benchmark found models often answer legal questions without asking for the missing facts.

Confirmed · 4 Manifests

The challengers post numbers while Nvidia changes chiefs

Cerebras reported record quarterly revenue of $210.8 million and raised its full-year outlook, AMD's MI430X won places in three supercomputers, Axelera says it has over $300 million in commitments for its Titania chip, and Exowatt launched new power hardware for AI data centres.

Confirmed · 3 Manifests

Most developers use agents. Few run them as a system.

Nine in ten professional developers now use AI coding agents at least weekly, but only about a quarter use automated agent workflows. The next adoption levels, one writer argues, are ecosystems of competing agent teams and simulations run repeatedly to filter out luck.

Research & Science

Finance & Investing

Health & Biotech

Sports

Federal Government

State Legislatures & Executives

Energy & Climate