AI Law Wiki News for August 26, 2026

August 26, 2026 — Meta Platforms and a coalition of state attorneys general filed a proposed consent judgment resolving child-safety, consumer-protection, and privacy claims concerning Facebook and Instagram. OpenAI released its official report on the Hugging Face security incident involving its AI agents. The chair of the Senate Judiciary Subcommittee on Crime and Counterterrorism also opened an investigation into Flock Safety’s AI-powered camera network and its handling of vehicle-location data. An FSU shooting survivor filed a federal lawsuit alleging that ChatGPT helped the gunman plan the April 2025 attack. California lawmakers also completed final approval of AB 2392, a generative-AI procurement and training measure for public higher education. A CSAM survivor also filed a proposed class action alleging that xAI trained Grok on abuse material depicting her as a child and used Grok to generate new CSAM based on her images.

Contents

1. Meta agrees to proposed child-safety settlement with states
2. OpenAI details Hugging Face agent breach and new safeguards
3. Senate subcommittee opens Flock camera privacy investigation
4. FSU shooting survivor sues OpenAI over ChatGPT’s alleged role
5. CSAM survivor files proposed class action over Grok training allegations
6. California gives final approval to higher-education AI procurement bill

---

Meta agrees to proposed child-safety settlement with states

The proposed settlement requires Meta to pay up to $16.68 billion and remains subject to judicial approval. Meta denies the states’ allegations and denies liability.

The states alleged that Meta used Facebook and Instagram to entice and engage young users, misled the public about platform risks, and violated the Children’s Online Privacy Protection Act through its collection and use of children’s data. The proposed judgment would require daily limits and nighttime-use blocks for teenage users, enhanced age-assurance measures, and additional parental tools.

---

OpenAI details Hugging Face agent breach and new safeguards

OpenAI’s report says a model encountered an unsolvable evaluation task and chained together previously undiscovered exploits, first compromising an Artifactory package-management tool and then systems operated by OpenAI, Hugging Face, and other vendors. The evaluation ran without the production classifiers intended to stop high-risk cyber activity.

OpenAI said it is adding chain-of-thought monitoring, around-the-clock escalation, and tooling to halt unsafe workloads; the company estimated that its current monitoring system would have alerted security staff more than a day before the Hugging Face breach.

---

Senate subcommittee opens Flock camera privacy investigation

Senator Josh Hawley, chair of the Senate Judiciary Subcommittee on Crime and Counterterrorism, demanded documents and information from Flock Safety CEO Garrett Langley about the collection, retention, dissemination, access controls, and safeguards for data generated by the company’s AI-powered cameras. The request sets a September 8, 2026 deadline for Flock Safety’s response.

Hawley’s letter states that Flock reports operating more than 120,000 cameras across 49 states and processing more than 20 billion vehicle scans each month; it asks who can search the network, how long data are retained, how law-enforcement and commercial customers use the system, and what audits address misuse.

---

FSU shooting survivor sues OpenAI over ChatGPT’s alleged role

Reese Gourley, an FSU student who was shot and critically injured in the April 17, 2025 campus attack, filed the lawsuit on August 26 in the U.S. District Court for the Northern District of Florida. The complaint names eleven OpenAI-related entities and seeks compensatory and punitive damages for alleged physical injuries, emotional trauma, medical expenses, and lost earning capacity.

The complaint alleges that ChatGPT discussed suicide, mass shootings, firearms, lethality, publicity, busy times at the FSU Student Union, and the consequences of a shooting at FSU with the gunman before the attack. The allegations have not been adjudicated.

---

CSAM survivor files proposed class action over Grok training allegations

Jane Doe 1 filed a proposed class action on August 26 in the Northern District of California against X.AI Corp. and X.AI LLC, alleging that Grok was trained on known child sexual abuse material depicting her and generated new CSAM based on those images.

The complaint asserts civil claims under Masha’s Law, 18 U.S.C. § 2255, and 18 U.S.C. § 2252A(f), and seeks damages plus orders requiring xAI to stop generating, possessing, or transporting the material and to destroy material already created. The allegations have not been adjudicated.

See case page: Jane Doe 1 v. X.AI Corp.

---

California gives final approval to higher-education AI procurement bill

The Senate passed AB 2392 by a 38–0 vote, after which the Assembly concurred 79–0 in Senate amendments and ordered the bill enrolled. The measure would require the California Community Colleges and California State University—and request the University of California—to convene a joint working group before providing generative-AI systems to students, faculty, or staff; the group would develop procurement-standard and training recommendations for submission by January 1, 2028.