US and AI: Trump's recipe for control may not be enough

Last week, US President Donald Trump convened the most influential CEOs of technology companies at the White House for a meeting to discuss and sign a "pact" on the "self-regulation" of artificial intelligence.

This was a reaction to several incidents in recent months. Their common feature was that the so-called AI agents got out of control and without human knowledge, crossed the set security restrictions, penetrated the Internet and carried out unauthorized cyberattacks.

Trump said the pact was a "morally binding" document that would serve as a "form of protection" against the potential risks of artificial intelligence. The companies that signed it are responsible for ensuring the security of their own technologies and have also agreed to put in place safeguards so that AI models can quickly detect any problems.

There was also an agreement to rename artificial intelligence to "superintelligence", which is said to better reflect its essence. Trump even signed an executive order ordering US government departments and agencies to start using the new name exclusively.

However, it cannot be overlooked that very critical voices have been raised in the US and elsewhere on the pact. It shows some kind of consensus, but it does not address the responsibility of AI developers, who should be held accountable for incidents. The non-binding nature of the pact negates the possibility of stricter regulation of the new technology and only creates room for alibis by companies that can only implement minimal protective measures against incidents.

Interestingly, coincidentally, in mid-September, a warning scenario from the AI Futures Project described a situation where artificial intelligence will completely get out of control. It states that the US president chose to "lightly regulate" it in 2029 because he considers warnings about the risks to be excessive. But as early as 2031, AI-controlled factories and armies of robots begin to operate autonomously in special economic zones. Then, of course, they begin to dominate the economy as human labor disappears. Artificial intelligence controls cars and trucks, planes and ships, power plants and internet hubs, and controls the media and politics. It is gradually taking control of the lives of everyone on the planet.

On the threshold of what may be a major technological leap since the Industrial Revolution, it is not entirely out of place to think more seriously about its potential impacts. Especially against the backdrop of today's geopolitically very turbulent times. But the White House message does not quite correspond to this.

Miloš Balabán, Comment in newspaper Mladá Fronta Dnes