AI Intelligence Digest
TIMPS
PostCards
AI in the Courtroom: Regulation Comes for Foundation Models
Issue 044 April 25, 2026 Hyderabad, India
⬡ Built with the TIMPS Ecosystem
Latest Issue
01 · LEAD
Legal & Policy

Musk v. OpenAI Trial Opens — The 'Terminator Warning' and the Battle for AI's Founding Story

Elon Musk took the stand for nearly two hours, calling himself 'a fool' for funding OpenAI and warning the jury about a 'Terminator outcome,' as OpenAI's counsel walked him through 2017-era exhibits showing he had explored a for-profit structure where he would hold majority equity.
Day 1 Musk testifies for 2 hours in OpenAI trial

The Elon Musk vs. OpenAI trial opened on April 27 in federal court before Judge Yvonne Gonzalez Rogers. Musk testified for nearly two hours, telling the jury he 'was a fool' for funding what became a startup, and warning about a 'Terminator outcome' from unregulated AGI development. OpenAI counsel William Savitt walked Musk through 2017-era exhibits showing Musk had explored a for-profit OpenAI where he would hold majority equity and control — directly contradicting his 'stole a charity' framing.

Microsoft sits as co-defendant on an aiding-and-abetting claim. The jury's verdict will be advisory; Judge Rogers will make the final call. The case has drawn intense interest as a referendum on whether OpenAI's transition from nonprofit to for-profit violated its founding mission. Day two featured OpenAI CEO Sam Altman's testimony, where he acknowledged tensions with Musk but defended the for-profit pivot as necessary to compete with Google and attract the capital needed for frontier AI development.

CNN, CNBC, The Verge
02 · CAPITAL
Regulation

EU AI Act's High-Risk Rules Become Enforceable — $35 Million Fines Loom

May 2026 marks the enforcement date for the EU AI Act's most consequential provisions: high-risk AI systems must now pass third-party conformity assessments, with fines up to 7% of global annual turnover or €35 million.
May 2026 EU AI Act high-risk rules enforceable

The European Union's AI Act entered its most significant enforcement phase, with rules for high-risk AI systems — including those used in critical infrastructure, education, employment, law enforcement, and healthcare — becoming fully enforceable. Companies deploying high-risk systems must maintain detailed technical documentation, implement risk management protocols, and pass third-party conformity assessments before market placement. Open-source model developers releasing general-purpose AI models must now provide training data summaries and energy consumption reports.

The enforcement represents the world's first comprehensive AI regulatory framework in action. Non-compliance carries fines of up to 7% of global annual turnover or €35 million. The EU AI Office has begun conducting investigations, with several unnamed US-based AI companies reportedly under preliminary review. Industry groups have criticized the cost of compliance, but the EU has signaled no intention to delay enforcement.

EU AI Office, Claude 5 Hub, AI Today
03 · DEFENSE
Defense & Policy

Google Signs Pentagon AI Deal With the Same Guardrails Anthropic Was Banned For

Google's contract includes language prohibiting domestic mass surveillance and autonomous weapons — almost identical to Anthropic's red lines — but explicitly does not give Google veto power, closing the circle on the dispute that led to Anthropic's blacklisting.
$200M Pentagon AI contract signed by Google

Google signed a classified AI contract with the Department of Defense on April 28, giving the Pentagon access to its AI for 'any lawful government purpose.' The contract includes language stating the AI 'is not intended for, and should not be used for, domestic mass surveillance or autonomous weapons (including target selection) without appropriate human oversight' — but explicitly does not give Google authority to veto operational decisions. That last clause is a direct response to the Anthropic standoff: the Pentagon designated Anthropic a supply-chain risk after the company refused to remove guardrails.

The contract closes a complicated chapter in Pentagon AI procurement. The DoD signed $200M agreements with Anthropic, OpenAI, and Google in 2025. After Anthropic's blacklisting, the Pentagon signed replacement deals with seven companies including Google, OpenAI, Microsoft, Amazon, Nvidia, SpaceX, and startup Reflection. Google's signing completes the circle — and confirms that the Pentagon's real dispute with Anthropic was about veto power, not guardrails.

Reuters, TechCrunch, The Information
04 · SIGNALS
5 Things Worth Knowing

5 Key Signals

01
GitHub Copilot moves to usage-based billing at 1 credit = $0.01; existing plans transition June 1
GitHub ↗
02
Anthropic launches Claude Security public beta — powered by Opus 4.7, scans enterprise codebases for memory corruption, injection flaws, auth bypasses
Anthropic ↗
03
CVE-2026-3854 disclosed: command-injection RCE on github.com via git push; 88% of GHES instances unpatched
Wiz, The Hacker News ↗
04
OpenAI's GPT-5.4-Cyber released on limited basis to vetted security vendors for vulnerability research
OpenAI, Economic Times ↗
05
Eightfold AI lawsuit over hiring discrimination proceeds as AI-in-HR cases become a new legal frontier
Towards Justice ↗
05 · THEMES
This Week in Focus

Top Themes

AI Legal Precedent
EU AI Act Enforcement
Pentagon AI Procurement
AI in HR Liability
Cyber AI Capabilities
Tool of the Week
Claude Security
Anthropic's new enterprise codebase scanner that finds high-severity vulnerabilities pattern-matching tools miss, with adversarial verification and auto-fix via Claude Code.