Since ChatGPT was released last year, new AI tools and services have captured people’s attention. Now, Meta warns that bad actors have noticed interest in AI chatbots. Facebook’s parent said scammers are creating ChatGPT malware and similar tools.
In a security report released Wednesday, Meta said it has discovered 10 families of malware posing as ChatGPT or related tools since March. Some malware, which can steal your personal information and compromise accounts, came in the form of browser extensions and links. Meta said it has removed more than 1,000 malicious links from its apps.
“The generative AI space is evolving rapidly and the bad guys know it,” said Jay Rosen, Meta’s chief information security officer. “As an industry, we’ve seen this across other popular topics of their time, such as cryptocurrency scams fueled by interest in digital currency.”
Rosen compared the use of attention in AI chatbots for malicious purposes to the cryptographic scams that flourished on social media just a few years ago.
Generative AI such as ChatGPT and Google Bard have exploded in popularity this year. Not only can users ask questions to these chatbots, but they can also request a poem, write a cover letter, or even write music. More technology companies are looking to integrate AI into their services or develop their own. This growing, almost insane interest in cryptocurrency for years makes those interested in artificial intelligence a prime target for scammers.
Rosen also noted that scammers hide across multiple services to avoid detection. This includes various social media platforms, different internet browsers, and file hosting services. If the scammers are caught on one platform, they can make minor modifications in order to continue spreading malware on another platform.
Meta said it will continue to roll out new protections against these malware campaigns and will work with other companies to stay on top of threats.
For more information, see CNET’s list of the best antivirus software of 2023 and 5 ways to secure your data after a hack.