This month's briefing covers the launch and federal suspension of Anthropic's Mythos 5, as well as various cybersecurity, safety, and fairness concerns addressed by the Trump administration, New York State, Illinois, Colorado, Connecticut, and Pope Leo XIV. Meanwhile, Sam Altman and OpenAI are being sued for a teenager's overdose death, Florida has sued OpenAI, and Elon Musk's claims against OpenAI have been tossed. Read on for a deeper dive into these and more key updates.
Regulatory, Legislative & Litigation Developments
Anthropic Expands, Then Suspends Access to Powerful Cybersecurity AI Models
On June 2, 2026, Anthropic announced that it was expanding Project Glasswing, its defensive cybersecurity initiative built around Claude Mythos Preview, to approximately 150 additional organizations in more than 15 countries. On June 9, the company launched Claude Mythos 5, an update to Mythos Preview for vetted partners, and Claude Fable 5, a Mythos-class model intended for broader use with added cybersecurity and biological safeguards. Three days later, Anthropic said the US government, citing national security authorities, issued an export-control directive requiring it to suspend access to Fable 5 and Mythos 5 by any foreign national, including foreign-national Anthropic employees. Anthropic said the directive forced it to disable both models for all customers to ensure compliance, though access to other Anthropic models was not affected. The company said it understood the government's concern to involve a potential jailbreak of Fable 5, but disputed the basis for the action, stating that the reported technique involved previously known, minor vulnerabilities and that other publicly available models could identify the same vulnerabilities.
President Trump Issues an Executive Order Addressing AI and Cybersecurity Threats
On June 2, 2026, the president signed an executive order addressing the issue of cybersecurity threats posed by certain frontier AI models (such as Anthropic's Mythos). The EO requires administration officials to establish an AI cybersecurity clearinghouse with industry leaders to proactively discover and fix vulnerabilities in software. It also mandates the creation of a classified process for determining whether an AI model is a covered frontier model and a voluntary framework for developers of such models. Under the framework, developers would have the opportunity (but not the obligation) to provide the federal government with prerelease access to covered frontier models and collaborate with the feds on the selection of trusted partners who would get early access to the models.
New York Department of Financial Services Warns of Heightened Cybersecurity Risks Associated with AI
On May 21, 2026, the New York State Department of Financial Services issued an advisory industry letter about heightened cybersecurity risks related to certain frontier AI models that can identify vulnerabilities and exploits in information systems with unprecedented speed and scale. These frontier AI models are not widely available yet, but the Department warns such models may be released soon and urges regulated entities to improve their security posture in advance of these threats. Alongside the advisory, the Department also issued new "Guidance on Measures Regulated Entities Should Consider in a Heightened Cybersecurity Threat Environment" to help regulated entities protect their data and information systems.
Illinois Passes AI Safety Audit Requirement
With bipartisan support, the Illinois General Assembly has passed SB 315, which would require frontier AI labs to have their safety practices independently audited by a third party — the first such requirement in the nation. Illinois Gov. JB Pritzker has announced he plans to sign the bill into law. The legislation goes beyond existing state AI laws in California and New York by requiring independent verification that AI companies are adhering to their own safety commitments, rather than relying on self-reporting. Illinois' approach may serve as a model for future federal AI safety legislation.
Colorado Governor Signs Scaled-Back AI Law after Federal Litigation
On May 14, 2026, Colorado Gov. Jared Polis signed SB 26-189, a substantially narrowed version of Colorado's AI law. The new statute replaces the original Colorado AI Act with language proposed by the state's AI working group in March (as covered in our May 1, 2026, AI briefing).
The new law, effective January 1, 2027, requires businesses using automated decision-making technology in consequential decisions — such as hiring, lending, and housing — to notify consumers before those decisions are made and to provide explanations and appeal rights when outcomes are adverse. However, enforcement remains on hold following Magistrate Judge Chung's April 27 order barring Colorado from enforcing alleged violations pending resolution of xAI's preliminary injunction motion. Attorney general rulemaking is also required before enforcement can proceed. While the compliance burden under SB 26-189 is substantially lighter than the original law, discrimination liability under existing state and federal statutes remains fully in effect.
Connecticut's AI Responsibility Act Brings New Rules for Developers and Businesses
On May 27, 2026, Connecticut Gov. Ned Lamont signed Public Act No. 26-15 (formerly SB 5) into law. This act, referred to as the Connecticut Artificial Intelligence Responsibility and Transparency Act, addresses a broad range of AI-related domains. Key provisions relate to (i) frontier model whistleblower protections, (ii) employment-related AI tools, (iii) AI companions, (iv) generative AI content provenance and watermarking, (v) a regulatory sandbox, (vi) AI subscription disclosures, and (vii) WARN Act disclosures. Notably, the Act vests exclusive enforcement authority under the Connecticut Unfair Trade Practices Act. Provisions of the Act begin taking effect on October 1, 2026, with additional provisions effective on January 1, 2027.
Illinois Delays Proposed Notice Rules for Use of AI in Employment Decisions
On May 15, the Illinois Department of Human Rights (IDHR) proposed new notice rules governing the use of AI in employment decisions. These rules would require employers to notify employees and applicants when AI is used to make employment-related decisions, including in recruitment advertising, hiring, promotion, and discipline. As proposed, the rules would require employers to post notices identifying how the AI tool is used and disclosing the product and vendor.
On June 2, 2026, IDHR announced that a public hearing regarding these proposed notice rules had been "temporarily postponed" to allow for "continued collaboration with other state agencies," signaling potential revisions to the proposal. IDHR has not announced a revised timeline for the public hearing.
Florida Sues OpenAI, Alleging Deceptive Practices
On June 1, 2026, the Florida attorney general filed a lawsuit in state court against OpenAI and its CEO, Sam Altman, alleging that ChatGPT was launched using deceptive practices, while safety warnings about potential harms were disregarded. In the complaint, Florida claims that ChatGPT's rollout has led to incidents of mass shootings, self-harm, and addiction, especially impacting minors. Florida seeks to hold OpenAI accountable under the Florida Deceptive and Unfair Trade Practices Act (FDUTPA), and aims to hold Altman personally liable for "reckless and willful conduct" as founder and CEO of OpenAI. Florida Attorney General James Uthmeier stated this is the first state-led lawsuit against OpenAI, asserting that the company and its CEO "ignored internal and external safety warnings, put children at great risk, and allowed a dangerous product to reach millions of Floridians."
Sam Altman and OpenAI Entities Sued for Role in Overdose Death
The parents of a 19-year-old man who died of an accidental drug overdose recently filed an action against the OpenAI Foundation, its corporate affiliates, and its CEO and co-founder Sam Altman. The complaint alleges that OpenAI's ChatGPT-4o model is a defectively designed product that caused their son's death and that OpenAI's and Altman's actions in developing and marketing it were negligent and in violation of California laws prohibiting unfair competition and the unauthorized practice of medicine. In addition to damages, they seek an injunction requiring OpenAI to add certain safeguards to ChatGPT-4o and to prevent the operation of OpenAI's ChatGPT Health, a product providing health care related guidance, pending evaluation by an independent third-party.
Court Tosses Musk's $130 Billion OpenAI Claims on Limitations Grounds
In what had the potential to be a blockbuster case, Elon Musk sued Sam Altman, Greg Brockman, and OpenAI in August 2024 in the US District Court for the Northern District of California, alleging that the defendants betrayed OpenAI's founding mission as a nonprofit dedicated to developing artificial intelligence for the benefit of humanity. Musk claimed that Altman and OpenAI accepted approximately $38 million in donations (contributed between December 2015 and May 2017) and then backed out on a promise to keep OpenAI a nonprofit organization in order to capitalize on the value of OpenAI. In the case, Musk sought in excess of $130 billion in damages, as well as a mandatory injunction that would have returned the company to a nonprofit organization.
At trial, key evidence focused on whether Musk brought his claims too late. Musk testified that, among other things, he did not learn OpenAI was operating as a for-profit company until 2022, when Microsoft was preparing to invest $10 billion in OpenAI. However, other evidence demonstrated that Musk himself advocated for transitioning OpenAI to a for-profit model as early as 2017, and that OpenAI created a capped-profit subsidiary in 2019. After a three-week trial, an advisory jury, after deliberating for less than two hours, found that all claims asserted by Musk were barred by the statute of limitations. As a result, the district court dismissed all claims asserted by Musk with prejudice, thereby completely avoiding the issue of whether Musk had actually been misled by Sam Altman and the other defendants. Musk has vowed to appeal.
Financial Stability Board Issues Consultation Report
The Financial Stability Board has released a consultation report on the responsible use of AI by financial institutions that's quite good. The report does not seek to establish an international standard or impose a prescriptive approach for responsible AI adoption. Instead, it provides thorough, helpful, and up-to-date guidance on such topics as agentic AI, monitoring and performance testing, human oversight, data governance, explainability/transparency, and third-party risk management.
Vatican Calls for Strong Oversight and Accountability in AI Design
On May 15, 2026, Pope Leo XIV signed his first encyclical, Magnifica Humanitas, a papal statement that strongly emphasizes the need for safeguarding human dignity in the age of artificial intelligence — the latest chapter in a decade-long Vatican dialogue with Silicon Valley on the human cost of AI. Pope Leo XIV declares that AI is "never a purely technical matter" and that when it affects employment, credit, public services, or reputation, there is a need to clearly define and assign responsibility at every stage of design, development, and deployment. The encyclical calls for robust legal frameworks, independent oversight, algorithmic transparency, and effective accountability mechanisms. It challenges the concentration of data, computing power, and platform control in private hands, asserting that data should be managed as a common good and that algorithmic power must be subject to public oversight. While not legally binding, experts in the tech industry and academia have already called the document a likely "benchmark" in AI governance debates.
Summer associates Lauren Bates, Allie Howard, Esther Huescas, Danielle Rossi, Petra Sikorski, and Natalie Song contributed to this briefing.
