The rapid advancement of AI technologies capable of creating persuasive text, images, and even live video is leading to increasingly sophisticated scams, according to cybersecurity experts speaking with AFP. They warn that internet users must be on high alert.Recent incidents, such as a notorious “romance scam” in France where a woman lost 830,000 euros (approximately $840,000) and fraudulent donation campaigns for Los Angeles fire victims, illustrate that anyone—be it individuals or businesses—can fall prey to cyberattacks, as noted by Arnaud Lemaire from cybersecurity firm F5.Phishing remains one of the most prevalent types of cyberattacks, where malefactors send deceptive emails, texts, or messages to trick users into taking harmful actions, such as clicking links, installing malware, or sharing sensitive information.
As Lemaire explained, AI chatbots using large language models (LLMs) help attackers craft more elaborate and time-efficient phishing messages, significantly reducing language clues that might expose a non-native speaker impersonating someone else.However, text generation is merely the surface of AI capabilities.
AI can leverage vast amounts of breached data to automate the production of highly personalized scams, a feat that would have required a large workforce only a few years ago, according to Steve Grobman, Chief Technical Officer at McAfee. Instead of quick hits, attackers often seek to build trust with specific individuals within targeted organizations over extended periods.If an employee is successfully deceived, attackers might bide their time until that person gains influence or an opportunity arises to extort funds, as highlighted by Martin Kraemer of cybersecurity training firm KnowBe4.
The dangers of such scams were starkly illustrated in February 2024, when scammers conned a multinational corporation in Hong Kong out of $26 million. In this case, a finance worker was led to believe he was having a video meeting with the company’s CEO and other executives, all of whom were actually AI-generated deepfakes.
The latest generation of deepfake technology has evolved to a point where most consumers struggle to distinguish between AI-generated images and authentic ones,” Grobman remarked.
To counter these threats, internet users should apply the same skepticism to video content that they often use for still images—where “photoshop” has become a common term. Simple measures, such as cross-referencing a purported news video with trusted sources, can help.In personal interactions, Lemaire humorously compared this caution to a BDSM dynamic, suggesting that users establish a “safe word” for verification. For instance, if a CEO requests a significant money transfer, adding a personal verification step can help ensure authenticity.
He also noted that asking the video caller to move their camera around may be an effective tactic for exposing AI deepfakes, as current AI struggles with such tasks.
PROF’LEE9JA⛱️ MEDIA
