Brief
Bill Gates Urges AI Safety Beyond Kill Switches
The Microsoft founder warned that AI kill switches alone are not enough, urging governments to enforce active tracking of models and robust safety measures.
By Felo News Desk · Published
Microsoft founder Bill Gates told NBC’s "Meet the Press" that emergency cut‑offs for artificial intelligence, the so‑called "kill switch", are not sufficient on their own. He said that internal monitoring and built‑in safety mechanisms are far more critical than shutdown tools, adding that governments should mandate active tracking of models’ behaviour within labs.
Gates said the most pressing threat comes from bad actors using current AI tools to carry out cyberattacks, scramble bank accounts or shut down power grids, rather than rogue autonomous machines. "Now, small groups with AI can do what only the biggest countries could do," he told the interview. "So it’s really the bad intent plus AI that looms as a major threat."
He dismissed concerns that AI safety rules would slow U.S. tech companies, replying to President Donald Trump’s stance that regulation would handicap American innovation in the race against China. Gates said the U.S. has a history of driving global consensus on safety standards, citing past agreements on aviation safety, environmental policy and nuclear arms.
Gates also referenced recent warnings from Anthropic whistleblower Jacob Coxon and industry executives such as Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman, who have called for coordinated safety standards and slower development. He noted that the U.S. and China had recently established a working dialogue to assess AI risks following a summit between Trump and Chinese President Xi Jinping.
"Regulating AI could prove harder than Cold War nuclear treaty negotiations," Gates said, stressing that the U.S. must act to prevent large negative outcomes from AI misuse.
Key facts
- Bill Gates said AI kill switches are not enough and called for internal monitoring (independent.co.uk)
- Gates warned that bad actors could use AI to conduct cyberattacks and disrupt critical infrastructure (independent.co.uk)
- Gates dismissed concerns that AI safety rules would slow U.S. tech companies (independent.co.uk)
- Gates referenced Anthropic whistleblower Jacob Coxon and executives Dario Amodei and Sam Altman as sharing similar safety concerns (independent.co.uk)
- Gates noted a recent U.S.–China working dialogue on AI risks after a summit between Trump and Xi (independent.co.uk)
Background
Earlier in the week, OpenAI CEO Sam Altman warned of catastrophic AI scenarios and called for federal regulation, while China rejected U.S. calls for AI regulation, stating it would not slow its AI development.
Why it matters
The debate highlights growing concerns over AI misuse and the need for international cooperation on safety standards.
What happens next
Gates called on Washington to mandate active tracking of AI models and for governments to insist on various safeguards, though no specific policy steps were announced.
Sources
- [1] independent.co.uk — originally reported as “Bill Gates warns AI kill switch is ‘not enough’ and says China isn’t greatest fear”
