Khaberni - Artificial Intelligence managed to uncover the identity of a group of fake accounts on social media platforms in the latest study conducted by the Swiss Federal Institute of Technology in Zurich in collaboration with the company "Anthropic" according to a report by the American tech site "The Verge".
The study relied on a number of artificial intelligence models that were not named; it included commercially available models from companies as well as open-source models that any user can install on their device, and all models were capable of accessing the internet and performing extensive searches on it.
These models were able to identify fake accounts through a variety of factors including information found in anonymous posts, as well as writing styles, posting frequency, and other aspects that can be deduced from the posts, then matching these aspects with millions of identified accounts on social media platforms to determine a group of similar accounts, then gradually narrowing down this list until it reaches the final account or group of accounts that may be linked to it.
The study showed that artificial intelligence revealed 68% of fake accounts with an accuracy of up to 90% in most cases, varying with the models used.
A separate report from the British newspaper "The Guardian" points to security concerns about the spread of this use of artificial intelligence models, as governments might use them to uncover the identities of opponents and activists via social media platforms, in addition to hackers using it to execute more precise phishing attacks.
The field of AI-enhanced surveillance is one that is rapidly evolving and raising concerns among computer scientists and privacy experts, as the technology can access information about users online in a way that was not possible before.
In this regard, Simon Lerman, the co-author of the study for "The Verge" site, explained that the total cost of conducting the experiment did not exceed two thousand dollars, with the cost per account being between one and four dollars.
This raises concerns for Peter Bentley, a professor of computer science at the University of London, who stated in his talk to The Guardian that this technology could provide commercially reliable products, which increases the risk of its use.
Also, the error rate in the results of artificial intelligence still exists and relying on this technology could cause erroneous connections between user accounts and anonymous posts found on social media platforms.
The study called for a series of measures to protect users' data from artificial intelligence models, including improving the encryption mechanisms on social media platforms and platforms that contain user information, in addition to limiting the amount of data that artificial intelligence models can access.
