Vitalik Buterin opposes the involvement of AI in financial management

Ethereum co-founder Бутерін" href="https://noworries.news/biography/vitalij-buterin/" data-bio-id="3642">Vitalik Buterin warned against the idea of transferring the authority to manage financial or organizational systems to artificial intelligence. According to him, the risks of such an approach far outweigh the potential benefits, as algorithms can be deceived and used for fraudulent purposes.

Buterin explained that if artificial intelligence is entrusted with making funding decisions, attackers will quickly learn how to manipulate it. Using the so-called jailbreak teams, they will be able to redirect resources into their own hands instead of fair distribution.

The discussion heated up after EdisonWatch CEO Eito Miyamura discovered a vulnerability in the new version of ChatGPT. The update allowed the service to interact with popular applications such as Gmail, Notion, and Google Calendar. Miyamura showed that this could be used to steal users' personal data.

According to his demonstration, an attacker can send an invite to the calendar with a hidden command that will be triggered when the user asks ChatGPT to prepare a daily schedule. In this case, the artificial intelligence can be hacked and give the hacker access to emails or other private information.

As an alternative to complete «machine control,» Buterin proposes to create a system of «infofinance,» a market where different AI models compete under constant supervision. According to his idea, independent reviewers and juries with the participation of humans and language models should act as controllers. According to Buterin, such a multi-level verification will help to quickly identify shortcomings and encourage developers to follow fair practices.