AI Law Wiki News for August 17, 2026

August 17, 2026 — The Motion Picture Association and ByteDance established a copyright-safeguard framework for ByteDance generative-AI products; the Justice Department's Antitrust Division was reported to be investigating Andreessen Horowitz over board roles at competing AI-data companies; Round Hill Music filed copyright lawsuits against Suno and Anthropic over alleged use of its songs for AI training; Minnesota defended its AI nudification ban against xAI's First Amendment challenge; and California AB 1883 advanced to a final Senate vote after the Senate ordered it to third reading on August 17.

Contents

1. MPA and ByteDance establish AI copyright-safeguard framework
2. Justice Department investigates Andreessen Horowitz board roles
3. Round Hill sues Suno and Anthropic over AI training
4. Minnesota defends AI nudification ban against xAI
5. California AB 1883 advances to final Senate vote

MPA and ByteDance establish AI copyright-safeguard framework

The agreement covers ByteDance image and video models and products including Seedance, Seedream, TikTok, the TikTok USDS Joint Venture, CapCut, and Dreamina. The parties said recent Seedream 5.0 Pro and Seedance 2.5 releases reflected strengthened intellectual-property protections, while the memorandum provides for continued collaboration as the technology evolves.

The agreement followed a February cease-and-desist letter in which the Motion Picture Association alleged that Seedance 2.0 generated unauthorized videos using studio-owned characters and imagery. The association said the memorandum was an important step toward protecting intellectual property in generative AI, but it did not publicly disclose the specific safeguards.

---

Justice Department investigates Andreessen Horowitz board roles

The Justice Department's Antitrust Division has been investigating whether Andreessen Horowitz partners improperly hold board seats at competing artificial-intelligence data companies, according to reports published August 17. The reported inquiry centers on Databricks, where Andreessen Horowitz co-founder Ben Horowitz is a director, and Fivetran, where firm partner Martin Casado serves on the board.

The investigation concerns Section 8 of the Clayton Act, which restricts interlocking directorates between competing corporations. The department had not made a final determination, and the inquiry could conclude without action.

---

Round Hill sues Suno and Anthropic over AI training

Round Hill Music filed copyright lawsuits against Suno and Anthropic, alleging that the companies used hundreds of the publisher's songs without permission to train artificial-intelligence models.

Round Hill identified 500 works in its initial allegations and said it could amend the cases to list 10,000 or more compositions, with claimed damages potentially exceeding $1 billion. The publisher's catalog includes songs such as the Goo Goo Dolls' “Iris,” Bonnie Tyler's “Total Eclipse of the Heart,” the Kinks' “Lola,” and Dio's “Holy Diver.”

---

Minnesota defends AI nudification ban against xAI

Minnesota Attorney General Keith Ellison urged a federal judge not to halt the state's AI nudification ban, arguing that the law regulates a technological tool rather than protected speech. xAI contends that the statute violates the First Amendment by restricting protected visual expression.

The law took effect August 1 after U.S. District Judge Donovan Frank denied xAI's emergency request for a temporary restraining order; a preliminary-injunction hearing was set for August 19.

---

California AB 1883 advances to final Senate vote

The California Senate ordered AB 1883 to third reading on August 17, placing the workplace-surveillance measure in position for a final Senate floor vote. The Assembly passed an earlier version 52–12 on May 27, and the Senate Appropriations Committee approved the amended measure 5–2 on August 13.

As amended, the bill would prohibit employers from using AI-enabled workplace-surveillance tools capable of recognizing or inferring a worker's emotional state or collecting neural data. It would authorize Labor Commissioner and public-prosecutor enforcement, permit affected workers or their representatives to seek damages, and allow penalties of up to $500 per violation; the text includes safety and specified federal-contract exceptions. A Sacramento Bee report described the proposal as a limit on AI workplace surveillance that analyzes emotional state or collects neural data.