Bill Gates Says AI Crossed 5 Danger Thresholds, Urges Monitoring and New Taxes

3 articles · Updated · MIT Technology Review · Aug 26

Bill Gates said current AI systems have already crossed five danger thresholds—bio, cyber, psychosocial, job-market and control—prompting him to publish a new essay and begin a broader public campaign.
The 70-year-old former Microsoft chief said guardrails and public debate have lagged badly as frontier models rapidly improved, arguing AI can now enable cyberattacks, design novel molecules and replace a large share of entry-level white-collar work.
On biosecurity, Gates said any model able to create new molecules should be monitored and kept from being copied into unregulated settings, calling bioterrorism roughly 50 times scarier and likelier than a natural pandemic.
He paired the warning with policy ideas including human-reserved jobs, robot taxes and token taxes to fund a stronger safety net, while saying society faces 10 to 20 years of turmoil before any AI-driven abundance.