AI Law Wiki News for September 3, 2026

September 3, 2026 — Federal lawmakers unveiled two artificial-intelligence safety proposals: Sen. Bernie Sanders and Rep. Greg Casar announced forthcoming legislation that would ban artificial superintelligence and pause advanced-AI development, while Reps. Josh Gottheimer and Mike Lawler unveiled a bill directing NIST to develop AI-agent security standards. A senior Pentagon official also said Anthropic remains designated a supply-chain risk for the department and defense industrial base, despite an August 27 injunction setting aside one designation. The District of Columbia Court of Appeals struck Deutsche Bank National Trust Company's brief after counsel acknowledged using Google's generative-AI search tool without verifying four nonexistent authorities. NHTSA also opened an audit of Tesla’s Cybercab safety certification after the company began deploying the autonomous vehicles in Austin.

Contents

1. Sanders and Casar announce proposed superintelligence ban
2. Gottheimer and Lawler unveil AI-agent security bill
3. Pentagon says Anthropic remains a supply-chain risk
4. D.C. appeals court strikes brief over AI-generated fake citations
5. NHTSA audits Tesla Cybercab safety certification
6. Pacific leaders endorse regional AI principles

---

Sanders and Casar announce proposed superintelligence ban

Sanders and Casar announced the forthcoming Ban Artificial Superintelligence Act on September 3; the sponsors said the proposal would permanently prohibit development and deployment of artificial superintelligence and temporarily pause advanced-AI development until a new federal regulator establishes safety rules and a model-review process.

The proposal would establish a Cabinet-level AI safety agency advised by an Artificial Intelligence Advisory Board, authorize monitoring of frontier systems and supervision of dangerous-capability removal, and direct the United States to pursue international agreements and export controls aimed at preventing artificial-superintelligence development worldwide. It would also authorize forced dissolution of entities and prison terms of up to 20 years for individuals who violate or circumvent the proposed restrictions.

The sponsors' September 3 announcement described the measure as forthcoming, and contemporaneous reporting said they planned to introduce it; neither cited source identified a bill number.

---

Gottheimer and Lawler unveil AI-agent security bill

Gottheimer and Lawler unveiled the Stop Rogue AI Act on September 3, proposing that the National Institute of Standards and Technology develop standards, guidelines, and best practices for secure deployment of AI agents within one year of enactment.

The proposed standards would address continuous verification of agent actions, security and reliability evaluations, tamper-resistant activity logs, and machine-readable inventories of agents operating on organizational networks. The standards would generally be voluntary, while federal contractors seeking new contracts would be encouraged to meet them and federal civilian agencies would apply them through their security programs in coordination with the Cybersecurity and Infrastructure Security Agency.

---

Pentagon says Anthropic remains a supply-chain risk

Under Secretary of Defense for Research and Engineering Emil Michael said on September 3 that Anthropic was still designated a supply-chain risk at the Pentagon and for the defense industrial base. His statement came one week after the U.S. District Court for the Northern District of California set aside the designation challenged there and permanently barred the government from giving effect to the challenged actions.

Anthropic is also pursuing a separate challenge in the U.S. Court of Appeals for the D.C. Circuit. Michael did not explain how his statement fit the district court injunction or whether it referred to the separate proceeding.

---

D.C. appeals court strikes brief over AI-generated fake citations

In Douglas v. Deutsche Bank National Trust Company, the District of Columbia Court of Appeals struck the bank's appellee brief on September 3 after the court found multiple citations it could not verify and counsel confirmed that four cited authorities did not exist. Counsel told the court that she had used Google's generative-AI search tool to locate authority and had not independently verified the citations before filing.

The court referred the matter to the Office of Disciplinary Counsel for any investigation it considers appropriate and sent the sanctions-authority question to its Rules Committee for possible clarification. The published order said lawyers can no longer credibly claim ignorance of generative AI's tendency to hallucinate authority, while emphasizing that the court was not discouraging responsible AI use in legal practice.

---

NHTSA audits Tesla Cybercab safety certification

The National Highway Traffic Safety Administration opened Audit Query AQ26002 on September 3 to examine the process and technical data Tesla used to certify about 1,000 Cybercab vehicles as compliant with federal motor-vehicle safety standards.

Tesla began commercial deployment of the autonomous two-seat vehicles in Austin on September 3; the Cybercab lacks permanently attached steering, braking, acceleration, and mirror controls. NHTSA said the audit would examine the basis for Tesla’s certification, including whether Tesla determined that some Federal Motor Vehicle Safety Standards did not apply to the vehicle.

The audit is a regulatory compliance inquiry, not a finding that Tesla violated federal safety standards.

---

Pacific leaders endorse regional AI principles

The September 3 communiqué of the 55th Pacific Islands Forum records leaders' endorsement of the Pacific Digital and Artificial Intelligence Technology Principles. It calls for human-centred AI, digital sovereignty, protection of Pacific languages and knowledge, transparency, safety and meaningful human oversight.

The principles are to inform a Pacific AI Governance Framework for consideration in 2027. The communiqué records a regional policy commitment, not adoption of that future framework. September 6 is the date of the PINA report; this entry uses the communiqué's September 3 event date.