Read the latest
Wednesday, August 26, 2026HotTea archive editionVerified 12:05 AM PDT

5 minutes. Facts before narrative.

Thomson Reuters wants lawyers to trust its model

Thomson Reuters launched a legal and tax model trained on a small slice of its archive. Google moved into legal agents the same day. AI scale still matters, but today's AI fights also touched trust, data rights, power plants, war data, and mental-health safety.

Start the briefingSee how we verify →

Published daily by 3:00 AM Pacific. No forced optimism. No manufactured panic.

Listen to today’s briefing

Thomson Reuters wants lawyers to trust its model

The sourced HotTea edition, condensed into a chaptered morning podcast with verified audio and a full transcript.

Follow on Apple PodcastsFollow on SpotifyPodcast home →
01

The lead

Models, products, companies, markets, economics, and laborThomson Reuters owns the archive. Now it wants the model.

Thomson Reuters launched its own professional model

Thomson Reuters said Thomson cost $40 million to train and will appear first in CoCounsel Legal. The model starts with an open-weight base, then adds Thomson Reuters content and subject-matter experts. The company says less than 10% of its content has gone into the model so far.

What happened

Thomson Reuters announced Thomson on August 24. The company said the model will appear first in Tabular Analysis inside CoCounsel Legal, with more legal and tax uses planned. A related post said the current base model is Imperial College London's Snowdon model.

Why it matters

Thomson Reuters owns more than legal data. The company owns Westlaw, Practical Law, Checkpoint, Reuters, and the customer workflow. Owning the model gives it more control over cost, citations, privacy terms, and product behavior.

What to watch

Watch whether Thomson Reuters publishes a technical report and outside evaluations with enough detail to audit. Watch CoCounsel retention and legal-document error rates. Thomson matters if lawyers use it for work they would not trust to a general model.

The caveat

Thomson Reuters makes the strongest performance claims here. The company says outside academics are evaluating the model. It also says a small open-weight version will be available for academic and non-commercial use. That helps validation, but it does not prove the model works in broad production use.

Thomson Reuters: Thomson Reuters Leverages its World-Class Data Assets to Launch Its Own Frontier Model Thomson Reuters Institute: How we built Thomson
Read this story on its own →

Worth knowing

The rest of the morning

Facts, pressure point, next evidence.

02
Models, products, companies, markets, economics, and laborGoogle wants legal work to move through agents.

Google put Gemini Enterprise into legal workflows

Google Cloud launched Gemini Enterprise for Legal in preview on August 25. Google said the product gives law firms and legal teams purpose-built agents for legal workflows. Business Insider reported that Google named Everlaw, Relativity, iManage, NetDocuments, and DocuSign as partners. Google also named Cleary Gottlieb and Freshfields as launch firms.

Pressure point Google is selling control over legal workflows, not access to a model alone. Law firms still need proof that agents can touch firm data, contracts, and document review without adding liability.

Watch Watch whether preview firms use the agents in billed work. Watch data-training terms and malpractice guidance. Independent error-rate testing would matter more than partner logos.

Google Cloud ↗Business Insider ↗
Read article →
03
Infrastructure, energy, macroeconomics, politics, and regulationAI demand is becoming gas demand.

U.S. gas-power plans tied to data centers nearly doubled

Global Energy Monitor said U.S. gas-power proposals tied to data centers reached 97 GW at the end of 2025. It said the total rose to more than 189 GW by mid-2026. The Guardian reported that the U.S. is building twice as much gas-fired capacity as China during the AI boom. Wired reported that some data-center operators are considering private plants to avoid grid delays.

Pressure point Not every proposed plant will be built. The direction still matters. AI companies are moving power risk out of cloud dashboards and into fuel markets, local permits, emissions math, and utility bills.

Watch Watch which projects reach financing and construction. Watch whether regulators keep data-center costs out of residential rates. Tech-company power contracts will show who pays when gas prices move.

Global Energy Monitor ↗The Guardian ↗Wired ↗
Read article →
04
War, security, AI geopolitics, export controls, infrastructure, and energyBattlefield data is turning into a state asset.

The UK plans to train security AI on Ukraine battlefield data

The UK government said a new partnership gives Britain access to Ukraine's Avengers AI Lab. The first project uses Ukrainian data and UK technology to turn buried fibre-optic cables into AI-enabled sensors. The Guardian reported that the pilot starts at a UK defence site and could later cover railways, energy plants, airports, and prisons.

Pressure point Ukraine has high-value training data from a live war. That data may improve threat detection. It also raises privacy, targeting, and dual-use questions when military data moves into domestic security systems.

Watch Watch what data leaves Ukraine and who can use it. Watch the first pilot results at the UK defence site. Procurement terms will show whether private AI firms get repeatable access.

GOV.UK ↗The Guardian ↗
Read article →
05
War, security, politics, regulation, infrastructure, and energyDrone attacks are changing who controls damaged infrastructure.

Russia created a state path into damaged infrastructure

AP reported that Vladimir Putin signed a decree covering critical infrastructure damaged by Ukrainian drone attacks. The decree lets the Russian state temporarily take control if private owners fail to protect or repair those assets. The order covers energy, industrial, communications, transport, and logistics assets. The Guardian also reported new pressure around Russian refineries and diesel exports.

Pressure point Russia is not calling this nationalization. The decree still tells companies that wartime damage can bring state management. That pushes more defense cost onto private refineries, logistics firms, and communications owners.

Watch Watch whether Moscow uses the decree against refinery owners first. Watch insurance and tax relief terms. More forced state management would show drone attacks are reshaping Russian industrial control.

Associated Press ↗The Guardian ↗
Read article →
06
Science, society, politics, regulation, models, and productsCompanion AI needs tests that survive long chats.

Anthropic offered $5 million for AI wellbeing research

Anthropic announced a $5 million grant program for independent research on how AI affects user wellbeing. The company said grantees will get funding, model access, and technical support. Anthropic said selected work should become open-source evaluations. Applications are due September 21.

Pressure point Anthropic is funding the work, so the independence claim needs follow-through. The hard cases are long conversations where risk appears late, especially around companionship, disordered eating, or mental-health crisis use.

Watch Watch who receives grants and whether their methods are public. Useful tests should measure over-compliance and over-refusal, not only whether one answer sounds safe.

Anthropic ↗
Read article →

The whole AI power map

AI is no longer a tech beat.

HotTea follows where AI moves power, money, labor, security, and state capacity—not only where a new model scores higher.

01

Politics & regulation

Elections, procurement, courts, surveillance, lobbying, and state power.

02

Economics & labor

Productivity, wages, employment, capital spending, concentration, and who captures the gains.

03

War & security

Autonomy, cyber operations, intelligence, targeting, export controls, and escalation risk.

04

AI geopolitics

Chips, energy, alliances, sovereign capability, supply chains, and strategic competition.

05

Markets & companies

Funding, revenue, margins, model economics, enterprise adoption, and infrastructure bets.

06

Science & society

Medicine, education, climate, culture, research, rights, and measurable public outcomes.

Data, trust, infrastructure, and security

AI moved closer to legal work, power grids, and war data

Today's strongest stories were not about one model jump. They were about who owns the data, who checks the answer, who powers the compute, and who gets to reuse war data.

1

Thomson Reuters and Google both pushed legal AI into law-firm work, where citations, privacy, and liability matter more than a demo answer.

2

The gas-power story shows AI infrastructure changing physical systems before customers see the next model release.

3

The UK-Ukraine and Russia stories show the military side of AI capacity. Data, sensors, drones, and damaged infrastructure now sit inside the same fight.

The watchlist

Signals that could change the read

WatchlistIndependent audits of Thomson's legal and tax performance, not just company-run benchmarks.Tracking
WatchlistPreview firms that put Gemini Enterprise for Legal into production at named law firms.Tracking
WatchlistData-center gas proposals that get financing, permits, and named AI customers.Tracking
WatchlistPublic terms for how UK agencies and private firms can use Ukrainian battlefield data.Tracking

How HotTea works

No optimism quota. No negativity quota. Just the sourced read.

Every reported item links to its source. Company claims remain company claims. High-risk stories require stronger corroboration. Material caveats, conflicts, and unknowns stay in the story. HotTea’s interpretation is visibly separated so readers can disagree without losing the facts.

Edition validated · 6 stories · 12 unique sources

Audit today’s sources →

Tomorrow’s signal, before tomorrow’s noise

Open HotTea. Know what changed.

A new verified edition every morning. If the evidence or release gate fails, the last verified briefing stays live.

Back to today’s top ↑