The glaring security risks with AI browser agents

For the vast majority of current users of the vast world of the internet, Google has always been the search engine par excellence, to the point that we are now accustomed to the verb “to google” - we no longer use the word “internet” to refer to the online search network, this world is increasingly being reduced to just “Google”. However, with the rise of artificial intelligence and the complete upheaval of the tech industry, the giant's hegemony could be waning.

Recently, both OpenAI and Perplexity have unveiled their own search engines, called Atlas and Comet respectively. The references to the exploration of the heavens are no coincidence, as enthusiasts of this technology often compare research in this field to the approach to the universe outside planet Earth.

These search engines will be accompanied by “AI agents”, built-in features that are supposedly able to complete tasks on behalf of the user, clicking on websites and filling out forms. Although the prospect may seem interesting, beneath the surface lies one of the most worrying risks of our time: data security.