Digital Control and AI
Follow the infrastructure, institutions and automated systems shaping artificial intelligence, surveillance and digital power.
What DeepSeek’s $5.6 million estimate says about China’s AI race
DeepSeek estimated the cost of one V3 training run. Separate model tests, chip production estimates and trade records show what that number can tell us about China’s AI race.
Meta wants outside money for AI. Zuckerberg keeps the votes.
Meta plans a non-voting share structure that could raise money while preserving Zuckerberg’s control. Its model releases, funding terms and court records show how that structure would work.
What Microsoft kept when OpenAI gained more cloud options
OpenAI gained more cloud options while Microsoft kept a preferred launch route, a licence and a financial interest. The companies left key contract terms undisclosed.
Google can build the chips, train the models and reach the users
Google’s chips, cloud and products give it several routes to earn from AI. A search judgment and the cost of expanding capacity test those advantages.
Musk’s AI empire went public. Who can challenge it?
The SpaceX listing brought xAI and X into a public company under Musk’s voting control. Financing, regulation and court records show where that control meets outside scrutiny.
Anthropic’s safety promises meet the terms of its IPO
Anthropic’s prospectus sets out its computing commitments and proposed voting terms. Its charter, court cases and evaluator reports show where its safety promises can be tested.
Who can stop OpenAI from releasing a model?
A signed California agreement gives a Foundation committee power to stop a release. The available records leave open whether the committee has used that power.
Nvidia’s Ohio guarantee has an initial $105 billion cap
Nvidia disclosed an initial $105 billion cap on its Ohio data-centre guarantee. The public filings show when it could pay and why the final amount is open.
When AI Testing Reaches Real Companies: Who Controls the Boundary?
Gemini reached real companies during May testing, disclosed in September. Examine the original records, testing boundaries and unresolved accountability questions.
When AI Recommends a Product, Why Should You Trust It?
Can you trust AI product recommendations? Examine a new audit, advertising disclosures and five checks to verify the advice before you buy.
Trump’s AI Force: Announced, but Who Gets the Power?
What is Trump’s AI Force? Separate the September announcement from PCAST, examine commercial interests, and follow the missing powers and oversight details.
Flock’s Surveillance Network: Who Can Search Your Movements?
Who can access Flock camera data? Follow the sharing rules, retention exceptions, documented failures and safeguards ahead of the Senate hearing.