AI Law Wiki News for April 30, 2026

AI law developments reported on April 30, with clearly dated follow-ups.

---

Musk v. Altman: April 30 testimony

On April 30, Musk testified that he knew of early for-profit discussions but had been reassured OpenAI would remain nonprofit. William Savitt questioned him about a 2017 term sheet and why he had not sued sooner. Musk also acknowledged xAI’s use of OpenAI models for training or validation. These were contested trial statements, not findings of liability. Judge Gonzalez Rogers limited testimony about AI safety; the Associated Press reported her observation that Musk was operating xAI in the same field.

See the case page for later orders.

Meta Threatens to Shut Down in New Mexico Over Child Safety Order

Meta filed a brief on April 30, 2026 warning that it may be forced to completely withdraw Facebook and Instagram from New Mexico if a state judge orders the company to adopt sweeping new child safety features. The threat comes in the second phase of a trial that already resulted in $375 million in civil penalties after a jury found Meta knowingly harmed children's mental health and concealed what it knew about child sexual exploitation on its platforms. The bench trial on remedies begins May 4, 2026.

Prosecutors seek court-ordered changes including effective age verification (99% accuracy), removal of addictive features like infinite scroll and auto-play, warning labels about platform risks, permanent bans for adults engaged in child exploitation, and an independent oversight committee. Meta called the demands "technically impractical, impossible for any company to meet and disregard the realities of the internet." New Mexico Attorney General Raúl Torrez responded that "Meta's refusal to follow the laws that protect our kids tells you everything you need to know about this company and the character of its leaders." The case is the first to reach trial among more than 40 state AG lawsuits against Meta over youth mental health.

---

Bipartisan CHATBOT Act Introduced in Senate

The CHATBOT Act was introduced on April 28 as S. 4407. See the April 28 digest for the sponsors’ proposal and official bill text.

---

White House Opposes Anthropic's Mythos Expansion

On April 30, The Next Web, relaying Bloomberg and Wall Street Journal reporting, said the White House opposed expanding Anthropic’s Mythos access to roughly 70 additional organizations because of security and computing-capacity concerns. This was press reporting about an objection, not a published legal prohibition. Anthropic’s own April 7 Project Glasswing announcement described limited access for defensive software-security work.

---

Hegseth Calls Anthropic CEO Amodei an "Ideological Lunatic" (April 30, 2026)

During the April 30 Senate Armed Services hearing, Pete Hegseth criticized Anthropic’s terms and called its leader an "ideological lunatic." Senator Jacky Rosen repeatedly asked about human control of lethal AI decisions. Hegseth said humans make decisions and ultimately said AI was not making lethal decisions. These were his representations to the committee.

The separate Mythos-access reporting is summarized above; Anthropic’s public announcement describes the defensive-security initiative.

See also: White House Opposes Anthropic Mythos Expansion

---

National-security AI agreements: subsequent May 1 announcement

In a subsequent May 1 release, the department announced AI agreements for classified networks with multiple companies and identified avoiding reliance on a single vendor as an objective. See May 1 news and the preceding Google reporting.