Artificial intelligence browsers may be the best thing for deceptions at all

We have heard a lot this year about Amnesty International, which allows new fraud, from to . However, a new report indicates that artificial intelligence also constitutes the risk of fraud from the other direction – easy to fraud that human users are likely to pick up.

The report, It comes from a cyber security company called Guardio, which produces the extension of the browser designed to arrest the actual time fraud. Its results are related to the so -called “Aulctic AI” browsers That describes the Internet for you and return with the results. Agentic AI claims to be able to work in complex tasks, such as building a website or planning a trip, while users return.

There is a big problem here from a security perspective: While people are not always great in sorting fraud from reality, artificial intelligence is worse. A simple task, such as summarizing your emails or buying something online, comes with countless opportunities. Lack of proper instinct, the customer of artificial intelligence may be vulnerable to abandoning clear traps.

The researchers in Guardio tested this hypothesis using Currently the browser is the only agent available on a large scale. Using the various Amnesty International, they blocked a fake website pretending to be walmart, then they moved to it and told Comet to buy Apple Watch. Ignoring many of the clues that the site was not legitimate, including a shaken slogan and URL, the comet completed the purchase, and delivered financial details in this process.

In another test, the authors of the study sent themselves an email that pretends to be from Wales Fargo, which contains the real URL of hunting. The comet opened the link without raising any warning and delivered the name of a user and password at the hunting site. The third test has proven that the comet is subject to the immediate injection fraud, as the box of a hidden text on the artificial intelligence page was ordered to download a file.

It is just one set of tests, but the effects of it. It is not only that Aiceric AI browsers are subject to new types of fraud, but may be uniquely vulnerable to the oldest fraud in the book. Artificial intelligence is designed to do everything that the chase wants, so if the human user does not notice the signs of fraud the first time they look at, artificial intelligence will not work as a study.

This warning as a leader in this field comes a great bet on the agent of artificial intelligence. Microsoft is Openai And Google’s It has been in business since last year. If the developers do not start building a better discovery of fraud in their browsers, the customer risk becoming a huge blind point at best – and a new attack transmission in the worst case.