AI Regulation & Policy

Today's Stories

1. 1

   UK forces Google to let publishers block AI training without search penalties

   The UK's Competition and Markets Authority ordered Google to provide an opt-out system that separates AI training rights from search display rights, covering Google AI Overviews and Gemini. Publishers can now block their content from being used to train AI while still appearing in regular search results, with main controls due by December 2026 and detailed page controls by March 2027.

   News websites and other publishers can now protect their content from being used by AI without losing Google search traffic, potentially changing how AI companies access training data.

2. 2

   Microsoft launches AI agent infrastructure at Build 2026 conference

   Microsoft announced Microsoft IQ as a context layer across its AI tools, along with Scout (a personal work assistant) and seven new AI models including MAI-Thinking-1. The company positioned itself as the platform for enterprise AI agents that can access company data securely and perform complex business tasks.

   Office workers may soon have AI assistants that can access company files and perform tasks like scheduling and data analysis more reliably than current chatbots.

3. 3

   Companies launch governance tools for AI agents handling financial tasks

   Devenex introduced governance infrastructure at Google Cloud Next 2026 to oversee AI agents that modify financial records and approve workflows. Separately, Kore AI's Artemis platform can now write, govern, and optimize AI agents from plain-language instructions with minimal human involvement.

   Businesses using AI to handle money and approvals will have better oversight tools, while the creation of business AI assistants becomes much faster and requires fewer technical experts.

4. 4

   Trump signs executive order requiring AI testing before public release

   After initially canceling the order, Trump signed an executive order on Tuesday requiring government testing of advanced AI models before companies can release them to the public. The order puts the NSA in charge of some testing processes and classifies certain evaluation methods.

   Future AI models like GPT-5 or Claude 5 will need government approval before launch, potentially slowing down new AI feature releases to consumers.

5. 5

   US closes chip export loophole affecting AMD and NVIDIA sales to China

   US regulators closed a loophole that allowed AMD and NVIDIA to sell advanced AI chips to Chinese customers through non-Chinese subsidiaries. The change directly affects both companies' international AI chip markets and introduces new supply chain compliance requirements.

   AI development in China may slow down due to reduced access to advanced chips, while AMD and NVIDIA face smaller international markets for their most powerful processors.

6. 6

   Senator Warren calls NVIDIA CEO to testify on China chip sales

   Senator Elizabeth Warren invited NVIDIA CEO Jensen Huang to testify on June 11 about the company's China business and compliance with US export controls on advanced AI chips. The request follows ongoing scrutiny of semiconductor exports amid geopolitical tensions.

   NVIDIA may face stricter oversight of its chip sales, potentially affecting the global supply of processors needed for AI development and training.