Global AI regulation intensifies: EU AI Act, White House safety tests, US sanctions debate

EU begins enforcing AI Act with fines up to €15M. White House finalizes voluntary safety tests with Meta, Google, OpenAI, Anthropic — classified benchmarking and 30-day pre-release review. US weighs sanctions on Chinese AI models, with potential $12B cost impact. Regulatory landscape tightening across jurisdictions. New: White House keeps AI evaluation framework secret — voluntary, not public. Decision undermines trust after OpenAI/Anthropic hacking incidents.