AI Regulation & Policy

Today's Stories

1. 1

   Illinois passes AI safety law requiring companies to test systems before release

   Illinois became the first U.S. state to require AI companies like OpenAI and Anthropic to conduct safety testing before releasing new systems. The law reduces President Trump's federal control over AI regulation and has support from major AI companies. It follows similar safety requirements being implemented in Europe and California.

   AI chatbots and tools used by Illinois residents may become safer as companies must now prove their systems won't cause harm before making them available to the public.

2. 2

   Major banks and tech companies launch ADG Framework for secure AI adoption

   Companies including Citi, JPMorgan Chase, Microsoft, and Salesforce created the Adopt. Defend. Govern. (ADG) Framework on May 28, 2026. The framework provides 12 security controls and governance guidelines to help organizations safely implement AI while complying with European AI Act regulations and U.S. standards.

   Businesses using AI tools for customer service, data analysis, and other operations will have clearer guidelines for protecting customer information and preventing AI system failures.

3. 3

   OpenAI releases Frontier Governance Framework aligned with EU and California regulations

   OpenAI published its internal safety and security practices on May 28, 2026, showing how the company aligns with emerging European Union and California AI regulations. The framework covers AI safety testing, security measures, and risk assessment practices for advanced AI systems.

   ChatGPT and other OpenAI tools may become more reliable and secure as the company follows stricter safety protocols required by multiple governments.

4. 4

   Companies face security challenges as AI agents gain autonomous decision-making power

   Enterprise software developers report growing concerns about AI agents (automated AI systems that can make decisions and trigger actions) accessing sensitive business systems without human oversight. Companies struggle to balance giving AI enough permissions to be useful while preventing unauthorized actions like editing databases or sending emails.

   AI assistants used in workplaces may become more restricted in what they can do automatically, requiring more human approval before taking actions that could affect business operations or customer data.

5. 5

   New UK privacy dataset released for training AI compliance assistants

   Researchers released a dataset of 5,000 question-and-answer pairs about UK privacy laws (GDPR) designed to help train AI systems that advise small businesses on data protection compliance. The dataset includes practical questions like consent requirements and specific legal references.

   Small businesses in the UK may soon have access to AI tools that can quickly answer questions about customer data privacy rules, making it easier to follow legal requirements.