News-May-07-2026

May 7, 2026 — Daily digest of AI law developments.

Contents

1. DOJ Files Formal Complaint in Intervention Against Colorado AI Law 2. Shivon Zilis Testifies in Musk v. Altman Trial 3. Anthropic Signs Compute Deal With SpaceX 4. Meta Asks Judge to Overturn Landmark Social Media Addiction Verdict 5. Google Chrome Silently Installs 4GB Gemini Nano AI Model Without Consent

---

DOJ Files Formal Complaint in Intervention Against Colorado AI Law

The U.S. Department of Justice filed a formal Complaint in Intervention on May 5, 2026, in xAI Corp v. Weiser, escalating the federal government's opposition to Colorado's SB 24-205 (the Colorado AI Act). The filing goes beyond the DOJ's earlier April 24 Statement of Interest, making the United States a formal party to the case. The DOJ argues that Colorado's algorithmic discrimination provisions violate the Equal Protection Clause and conflict with federal AI policy. Separately, a federal judge on April 27 paused enforcement of SB 24-205 pending further proceedings.

See full article: May 5, 2026 — DOJ Files Complaint in Intervention Against Colorado AI Law

---

Shivon Zilis Testifies in Musk v. Altman Trial

Shivon Zilis, former OpenAI board member and the mother of four of Elon Musk's children, testified for hours on May 6, 2026, in the ongoing Musk v. Altman trial in Oakland. Zilis pushed back on OpenAI's characterization of her relationship with Musk as a secret allegiance, testifying that her relationship with Musk did not influence her duties as a board member. She confirmed that Musk had offered her the opportunity to use his sperm for conception. Zilis left the OpenAI board in 2023 after Musk started xAI. Her testimony came on the same day that former OpenAI CTO Mira Murati testified that Sam Altman "sowed chaos and distrust" among senior leadership.

See full article: May 6, 2026 — Musk v. Altman Trial Day 6: Zilis and Murati Testify

---

Anthropic Signs Compute Deal With SpaceX

Anthropic announced on May 7, 2026 that it has signed an agreement with SpaceX to access all of the compute capacity at Colossus 1, one of the world's largest AI supercomputers. The deal provides Anthropic with over 300 MW of capacity and more than 220,000 Nvidia GPUs, substantially expanding the company's AI inference capabilities. In conjunction with the deal, Anthropic doubled Claude Code's five-hour rate limits for paid plans and removed peak-hours reductions for Pro and Max plans. The partnership comes amid Anthropic's ongoing dispute with the U.S. government over a supply-chain risk designation, and raises questions about the competitive dynamics between Musk's xAI/SpaceX and Anthropic, which has positioned itself as an AI safety-focused alternative.

---

Meta Asks Judge to Overturn Landmark Social Media Addiction Verdict

Meta Platforms asked a Los Angeles judge on May 6, 2026, to throw out the landmark social media addiction verdict or order a new trial, arguing that Section 230 of the Communications Decency Act shields it from liability and that the jury's finding was based on the content the plaintiff viewed rather than platform design features. The March 2026 jury verdict had found Meta liable for $4.2 million and Google for $1.8 million for negligently designing platforms that harmed a young woman's mental health. Google has also asked the court to set aside the verdict and plans to appeal. The case is the first bellwether trial in a coordinated proceeding involving thousands of similar lawsuits, and the Section 230 defense could reach the Ninth Circuit and potentially the Supreme Court.

See full article: May 6, 2026 — Meta Asks Judge to Overturn Social Media Addiction Verdict

---

Google Chrome Silently Installs 4GB Gemini Nano AI Model Without Consent

Privacy researchers revealed that Google Chrome has been silently downloading a 4 GB AI model (Gemini Nano) to hundreds of millions of users' devices without consent, notice, or functional opt-out. Forensic analysis by privacy researcher Alexander Hanff confirmed that Chrome profiles accumulated the model within 14 minutes of creation with zero human interaction, and that Chrome automatically re-downloads the model if deleted. Chrome 147's "AI Mode" omnibox pill creates a misleading impression that queries are processed on-device, when in fact AI Mode sends all queries to Google's cloud servers. The installation may violate the EU GDPR and ePrivacy Directive, and a German court ruling in March 2025 established precedent that browser-installed code requires informed user consent.

See full article: May 7, 2026 — Google Chrome Silently Installs 4GB Gemini Nano Without Consent

---