Bill Gates warns that the biggest threat from artificial intelligence is its malicious use, not the loss of control

The immediate danger is misuse, not autonomy

In an interview with NBC’s Meet the Press program, which aired on Sunday the 27th, Bill Gates said that the most pressing problem with artificial intelligence is the actions of people with bad intentions. He described scenarios such as fraud, disruption of power grids, manipulation of bank accounts, and even the development of biological weapons with the help of AI.

The Microsoft co-founder believes that the technology can already be considered an “intelligent species” that, in many ways, surpasses humans. However, he argues that AI does not yet control millions of robots or other physical devices, which means that if it becomes dangerous, the systems can be shut down.

Monitoring and safeguards as a solution

Gates argues that, instead of an emergency button or halting progress, the best protection is the implementation of monitoring and safeguard mechanisms directly into AI models. He warns that without these measures, police and authorities may be unable to identify fraudulent activities or cyberattacks.

“The same monitoring we put in place to prevent bad people from using AI is exactly what you want to see if AI itself is doing it,” he said.

He also highlighted the danger of biological applications, where the line between creating a new drug and developing a bioterrorism tool is very thin. “We need to monitor any sophisticated model and record everything,” he said.

For Gates, self-regulation is not enough. He calls on Congress to pass laws requiring these mechanisms, arguing that the additional cost to industry would be modest and would not impede technological progress.

View contrasts with Trump’s

Gates’ position contrasts with that of former President Donald Trump, who this month rejected any federal regulation of AI companies. The Microsoft co-founder notes that, unlike rockets and nuclear weapons, AI was developed almost entirely by private initiative, and governments still struggle to understand it.

He warns that treating the issue as a race between nations leads some to advocate “going full speed ahead.” “It is completely irresponsible not to require that all AI have safeguards and monitoring capabilities,” he said.

Frequently Asked Questions

• What is the biggest risk of AI according to Bill Gates? He says the most urgent danger is the malicious use of the technology by people with bad intentions, including fraud, attacks on power grids, manipulation of bank accounts, and the development of biological weapons.
• Does Gates think AI could get out of control? He considers it possible that future systems will become more powerful and act against human interests, but does not believe that AIs are currently out of control.
• Could AI destroy humanity? Gates does not rule out this scenario. He says AI is powerful enough to cause events that could result in one billion deaths.
• Why is the immediate risk not AI acting on its own? Because current systems do not control robots or other physical devices, it would still be possible to shut down the computers responsible for their operation if it became dangerous.
• What kind of attack does Gates fear people could carry out with AI? He cited fraud, disruption of electrical systems, manipulation of bank accounts, and the use of advanced models to create bioterrorism tools.
• What does Gates advocate to control the risks of AI? He calls for safeguards, continuous monitoring, and recording of activities, as well as legislation that requires companies to implement these mechanisms.
• Is Gates in favor of slowing down AI development? He would support a global plan to halt the advancement of the technology, but doubts that this will happen due to economic and geopolitical incentives.
• Why is the use of AI in biological weapons particularly concerning? Because it is difficult to distinguish legitimate applications, such as the creation of new drugs, from malicious uses for bioterrorism, reinforcing the need for monitoring.