Meta sees everything, Copy Fail, and a deepfake gets hired

In this episode of Smashing Security, Graham Cluley and guest Paul Ducklin dive into three major cybersecurity stories. First, they expose the dark reality behind Meta's smart glasses, revealing that thousands of workers in Nairobi, Kenya, are manually labeling AI training data from the glasses' recordings—capturing intimate moments like bathroom visits, credit card details, and private conversations—despite Meta's claims of privacy and user control. The workers, many of whom were traumatized by previous Facebook moderation work, were fired after whistleblowing, raising serious ethical and privacy concerns. Next, the hosts dissect the 'Copy Fail' Linux kernel vulnerability, a legitimate but overhyped bug with a catchy name that doesn't pose a critical threat like Heartbleed, but still warrants patching for privilege escalation risks. Finally, Jake Moore of ESET shares a chilling real-world deepfake experiment where he successfully impersonated a candidate in a job interview using AI, securing a £38,000 offer—highlighting how easily AI can bypass hiring processes and the urgent need for verification methods like phone calls and in-person meetings. The episode underscores the hidden human cost of AI, the dangers of overhyping bugs, and the growing threat of AI-powered social engineering.