Bill Gates says we’ve passed AI’s danger thresholds. Now what?

Bill Gates argues that AI has crossed danger thresholds in biological design, cyberattack, psychosocial dependence, white-collar substitution, and possibly control. He calls biological risk especially urgent, saying models capable of creating novel molecules should be monitored and proposing U.S.–China cooperation on that threshold.

His economic proposals include taxes on robots or AI tokens to fund a stronger safety net and reserving selected jobs for people, permanently or during a transition. Gates still expects benefits in health, agriculture, education, and navigating bureaucracy, but forecasts a turbulent transition before any broad abundance.

The interview combines concrete policy prompts with sweeping forecasts. Gates says well-defined white-collar work can be performed more cheaply by properly implemented AI and suggests almost every entry-level job has crossed that threshold. Those are forceful judgments, not labor-market measurements supplied in the article, and should be tested against deployment costs, error handling, organizational change, and demand effects.

His strongest policy instinct is that industry cannot self-regulate. Voluntary reviews are structurally weak when companies face incentives to advance and competitors operate globally. Monitoring molecule-generating models and improving government expertise are at least governable proposals, although the interview does not specify enforcement or how to distinguish beneficial research from dangerous use.

Signal Desk's read: Gates is right that threshold-based policy should begin before harms become undeniable, but token taxes and human-reserved jobs are sketches, not programs. Taxing usage could discourage beneficial applications alongside substitution, while job reservations could preserve dignity or freeze inefficient arrangements depending on design.

The useful contribution is forcing distribution into the AI debate. If cognition and later physical labor become much cheaper, corporate profit taxes, targeted excise taxes, shorter workweeks, transitional protection, and service provision all deserve comparison. The central question is who captures productivity gains and who bears the adjustment—not whether technology eventually creates abundance.

Gates is the former Microsoft chief executive and now a philanthropist whose foundation uses AI in vaccine and drug work. He plans additional essays, including one focused on biological risk, and is speaking with leaders in industry, government, and civil society.

He distinguishes a transition of roughly 10 to 20 years from a later possible steady state of abundance. He rejects universal basic income as currently unaffordable and instead raises targeted revenue and labor protections, while acknowledging that countries may choose different categories of human-reserved work.