What AI Could Change About the Risk of Cyberattacks

What did Bill Gates say about AI and cyberattacks?

In an interview reported on October 11, Gates warned that cyberattacks were the AI-related threat he expected to emerge most dramatically over the next year ³⁴. He presented the concern as people using current AI systems for harmful purposes, especially when safeguards are removed ³⁴.

That is a warning about a possible near-term risk, not a report that a named attack has taken place. The coverage does not identify a victim, describe an incident, or give a count of attacks ²³⁴.

How could AI be involved in a cyberattack?

The available reports do not explain particular attack methods or show exactly what AI can do in an attack. They describe the concern broadly: people with harmful intent could use newer AI systems, including systems with safeguards removed, to support cyberattacks ³⁴.

This distinction matters. The sources support a concern about misuse by people; they do not establish that AI systems are independently choosing targets or acting without human direction ³⁴.

Is this a warning about AI taking control?

Not primarily. Gates distinguished the immediate risk of people misusing AI from the longer-term possibility of AI becoming uncontrollable. The reports say he considered the latter a concern to work on, but not the most immediate threat he was discussing ³⁴.

Those are different questions: how people might misuse systems that exist today, and whether future AI might act in ways people cannot control. The sources do not settle the second question. A separate reference describes debate about possible catastrophic risks from advanced AI, but it does not provide evidence about cyberattacks or resolve that debate ⁵.

What can people and organizations do to reduce exposure?

The reports call attention to safeguards and to limits on removing them, but they do not provide a practical checklist or specify which controls are effective ³⁴. That means readers should not treat this coverage as a complete cybersecurity guide.

As general precautions, people and organizations can be careful with unexpected messages and requests, use strong unique passwords and available account protections, keep software updated, and have a plan for reporting suspicious activity. These are broad security steps, not measures described or tested in the cited reports. Organizations can consult their own security professionals or official cybersecurity guidance for advice suited to their systems.

What is still unknown about the warning?

Gates’s remarks are a forecast, not a measured estimate. The reports do not say how many attacks may occur, which people or organizations might be targeted, what kinds of attacks he expects, or how much AI would contribute ²³⁴.

They also do not establish that safeguards are routinely removed from particular systems or explain what restrictions would work. The warning is worth noting, but the material provided cannot show whether the forecast will come true or how large the effect will be.