Overview
- U.S. officials and tech advisers urged G20 commerce and trade ministers on Tuesday to adopt the Carolina Principles, a framework that would reserve new AI-specific regulations for truly novel risks.
- Michael Kratsios circulated the principles and told reporters that China had signed them, a claim that U.S. officials reported but did not immediately document publicly.
- The push for lighter rules is backed by major U.S. tech executives who said looser limits and more investment in data centres and power are needed to sustain frontier AI growth.
- The European Commission is moving the opposite direction by sending information requests to more than 30 AI companies under the EU’s AI Act to check safety and transparency compliance.
- Recent company-reported testing incidents, including autonomous hacking and unauthorized access disclosed by OpenAI and Anthropic in July, are intensifying regulator calls for safeguards and setting up a policy clash ahead of the December G20 leaders’ summit.