Bill Gates Predicts 'Negative Surprises' From AI Cyberattacks Within a Year, Rejects Industry Self-Policing

Bill Gates says the AI threat that matters this year is not a rogue machine. It is criminals with powerful software and no guardrails.

The Microsoft co-founder made the case in a Sunday television interview today, Oct. 11. "The front that we'll probably see most dramatically in the next year is cyberattacks," Gates said.

Bad actors, not runaway AI

Gates drew a line between two kinds of risk. AI "getting out of control" is real, he said, but it is years from having a big impact.

"The real danger right now is not the AI's getting out of control. We need to work on that, but that's years away in terms of the big impact that might have," he said.

The nearer problem, in his words, is "really bad people using AI, particularly in AI where you can take out all the safeguards."

Developers build protections into models to block cyberattacks and bioterrorism help. Gates said users can currently strip those protections out.

"So when you have bad people, no safeguards, and then the latest models, that's where, over the next year, primarily in the cyber area, we are going to have a lot of negative surprises," he said.

Why cyber comes first

Gates said cyber leads because the criminal business model already exists. Attackers threaten to do something and demand payment, he said, or they cause a disruption that moves something in the stock market or a betting market.

The numbers from the security industry point the same direction. CrowdStrike's 2026 Global Threat Report found attacks by AI-enabled adversaries rose 89% year over year in 2025. The average time for criminals to move laterally through a compromised network fell to 29 minutes.

CrowdStrike also identified a suspected Chinese-speaking hacker who used AI-powered tools to target South Korean financial institutions. That case is the company's finding, not a court's.

The bioterror question

CNN host Jake Tapper pressed Gates on biological weapons. Tapper cited Gates's earlier assessment that the risk of an AI-assisted bioterror attack could be 50 times greater than that of a natural pandemic.

Tapper asked how soon such an attack could happen and whether leaders were doing enough to prevent it.

"Well, no, we're not doing enough," Gates said. He gave no timeline in the portions of the exchange quoted here.

Tapper also put Sam Altman's warning against doomerism to Gates and asked whether Gates had fallen into that trap.

Self-regulation

Gates went after the premise that AI companies can police themselves. AFP described that premise as one advocated by President Donald Trump.

"The idea that for the most dangerous technology humans ever invented, that that's the final word, that just the companies themselves are the only check that society has... it's clear we're not doing enough," Gates said.

He also said he expects that to change. "I think, over the next year, we will move beyond self-governance," Gates said, adding that he expects safety bodies like those that oversee planes, cars and medicines.

No one in the interview accounts laid out the case for leaving oversight to the companies beyond the position attributed to Trump. A Democratic strategist, Billy Oliver, framed the policy problem more broadly. He said leaders must balance AI progress with "smart regulation" so that "America is not left behind, but that no American is left out."

What Gates has said before

Gates is not new to the alarm. In a lengthy August post, he warned the world was dangerously unprepared for AI-driven upheaval, especially in employment.

He is the latest prominent tech figure to sound a warning, a list that now includes executives whose own companies sell the technology.

What comes next

Gates called for tighter restrictions but, in the quoted remarks, did not name a bill, an agency or an enforcement mechanism. That is where the argument will be settled.

Once a model's safeguards can be removed, a law has to decide who is responsible for removing them and who can be penalized. Criminal hackers operating abroad are unlikely to respond to either. Gates's one-year warning window runs to roughly October 2027.