Israel
National Digital Agency Publishes Public Sector Responsible AI Guide
On May 19, 2026, the Israel National Digital Agency, in collaboration with the Ministry of Justice and the Ministry of Innovation, Science and Technology, published version 1.0 of its Guide for Responsible AI Use in the Public Sector, covering AI governance, risk management and guidance for end users. While the Guide is framed as general best-practice guidance rather than a standalone legal instrument, it is intended to guide public-sector organizations, and their employees, in procuring, developing, deploying and using AI systems through internal governance, risk assessment and mitigation processes. The Guide introduces a “traffic light” system for AI system approval and implementation processes based on risk levels, alongside a dedicated “blue track” for limited and supervised AI experimentation.
Israeli Government Approves National AI Program
On June 16, 2026, the Israeli government approved a national plan to accelerate AI development and adoption. The program aims to position Israel as a global leader by focusing on national computing infrastructure, research and development, workforce training, public sector adoption, and international collaboration. Key initiatives include establishing a national AI institute, creating “acceleration hubs”, promoting Cyber AI and Physical AI capabilities devising a long-term policy to address expected workforce changes resulting from the AI revolution.
USA
Colorado Scales Back AI Law
On May 14, 2026, the Colorado Governor signed SB 26-189, repealing the state’s earlier AI law (which was covered in our previous updates). The new law replaces the prior regulation of “high-risk” AI systems with a broader framework governing “automated decision-making technology” (ADMT) used in consequential decisions across key services or domains. In practice, it scales back the prior law’s anti-discrimination requirements and instead emphasizes notice, transparency and certain disclosures when covered systems are used to take adverse decisions about individuals.
The law takes effect on January 2027 and applies to consequential decisions made on or after that date.
Connecticut Enacts Law Targeting AI Companions and Automated Employment Tools
In May 2026, Connecticut enacted the Online Safety Act, introducing rules for AI companions (e.g., AI tools designed to sustain ongoing relationships with users), automated technologies used in employment and recruitment decisions, and certain AI-generated content. The act requires operators of AI companions to disclose when users are interacting with AI and implement safeguards relating to self-harm and minors. It also requires notice to employees and applicants when using automated tools that materially influence hiring and employment decisions, clarifies that employers cannot rely on such tools as a defense to discrimination claims and requires certain consumer-facing AI providers to embed provenance data in AI-generated or altered audio, image, and video content where commercially and technically reasonable.
This development signals rising compliance expectations for AI developers, employers, and consumer-facing platforms, particularly in high-risk use cases.
New York Legislature Passes Act Requiring the Disclosure of AI-Generated News Content
On June 8, 2026, the New York State Legislature passed the FAIR News Act, which would require news organizations operating in New York to provide clear and conspicuous disclaimers when published content is substantially or wholly generated by AI and to ensure such content undergoes human review prior to publication.
If enacted, the bill would be the first of its kind in the United States, marking a notable development in state-level AI regulation, and may encourage similar state initiatives to establish transparency requirements for AI use in media.
President Trump Issues Executive Order Advancing AI Model Security; Claude Model Suspension Immediately Follows
On June 2, 2026, President Trump issued an executive order titled “Promoting Advanced Artificial Intelligence Innovation and Security”. The order directs Federal agencies to enhance cyber security using AI tools and establishes a voluntary framework allowing AI developers to share frontier models with the government before they are released to the public to assess their cyber capabilities.
This order was published amidst growing concern regarding AI models exhibiting advanced cyber capabilities posing risks to national security. Merely 10 days following its release, Anthropic was ordered by the administration to suspend access to its Fable 5 and Mythos 5 models, citing national security concerns.
EU
European Commission Issues Draft AI Act Transparency Guidelines
In May 2026, the European Commission released draft guidelines clarifying how to apply Article 50 of the EU AI Act. This Article, effective starting August 2, 2026, imposes transparency and labeling obligations on providers and deployers of AI systems. Notably, the guidelines outline practical examples clarifying when providers of AI systems must inform users that they are interacting with an AI system, and the form of disclosure needed to maintain compliance.
EU Commission Issues Guidelines the Classification of High-Risk AI Systems
On May 19, 2025, the EU Commission published three draft guidelines, clarifying how to apply each of the 3 High-Risk AI system classification tracks under the AI Act (which subject covered AI systems to strict obligations under the AI Act). These guidelines notably provide essential clarity on the primary classification track for AI systems used in sensitive areas such as employment and essential services, which are presumed to be High-Risk, and the scope of exclusions for systems merely performing technical, preparatory or narrow, procedural tasks.
Critically, the guidelines clarify that implementing human oversight does not automatically exempt an AI system from High-Risk classification. Furthermore, they emphasize that these High-Risk exclusions should be interpreted narrowly, and provide examples to assist users in applying them.
Munich Court Imposes Liability on AI-Generated Search Summaries
On May 28, 2026, the Regional Court of Munich issued a landmark ruling concerning AI provider liability for defamatory AI-generated summaries. The case involved a search engine operator whose AI produced a summary containing defamatory content. The court held the provider liable for the result, finding that the AI-generated summary constituted the provider’s own content since it actively summarized and evaluated information, rather than just reproduce existing content.
While this ruling does not constitute a binding precedent, it encourages AI providers and deployers to implement guardrails and oversight mechanisms for AI outputs to mitigate potential liability.
The above content is a summary provided for informational purposes only and does not constitute legal advice. It should not be relied upon without obtaining further professional legal counsel.
