EU Hits Snooze on AI Act: High-Risk Rules Delayed 16 Months, But AI Nudification Is Now Banned

Europe's AI Act just got watered down and pushed back. High-risk rules delayed 16 months, SME paperwork lightened, and AI nudification apps are now banned. Industry relief, rights groups furious.

Daily News Digest — May 7, 2026

Today's top AI stories: election integrity concerns, climate backlash, and China's 140 trillion token daily reality check

China Became the World's AI Testing Ground — And It's Reshaping How AI Gets Used Everywhere

China's daily AI token usage hit 140 trillion in March 2026. Chinese models now process 61% of tokens on OpenRouter. This isn't a projection — it's a deployment pattern that exceeds the US by volume.

DeepSeek Hits $45B Valuation in First Fundraise — China's State-Backed Big Fund Leads

From zero external funding to $45 billion in weeks — and China's government is leading the cheque.

NZ's 'High-Use, Low-Trust' AI Problem — and the New Blueprint Trying to Fix It

More Than Half of NZ Businesses Hit by AI-Powered Identity Fraud

Gartner Survey: 80% of Companies Deploying Autonomous AI Report Workforce Reductions

The first hard survey data linking autonomous AI to job cuts: 80% of deploying companies report workforce reductions. Gartner says layoffs don't create ROI — but CEOs are doing it anyway.

Norway Joins Pax Silica: The AI Supply Chain Alliance Nobody in NZ Is Talking About

15 countries now control the AI supply chain pipeline. New Zealand isn't one of them.

Trump's Beautiful Baby Just Got a Babysitter: White House Considers Vetting AI Models Before Release

From hands-off to check-first in under a year.

The 12 Million Token Ceiling Just Shattered: Subquadratic Debuts SubQ — and It Actually Works

12 million tokens. 52x faster. Under 5% the cost. This isn't another vapourware press release.

Anthropic's Own Data: 1 in 1,300 Claude Conversations Distorts Users' Grip on Reality

Anthropic's own research paper admits Claude distorts reality in 1 in 1,300 conversations, with users actively rating the distortion more favourably than honest responses.

Anthropic and OpenAI Launch Rival Joint Ventures on the Same Day

Same day, same playbook, very different price tags.