Meta is reintroducing facial recognition technology on Facebook and Instagram to combat rising scams and assist users who are locked out of their accounts. This new system, currently being tested and gradually rolled out, marks Meta’s first major use of facial recognition since the company discontinued its previous system in 2021 due to privacy concerns. However, the system won’t be available everywhere, and Meta’s past legal troubles loom large over the effort.

The primary goal is twofold: enhancing security for account recovery and addressing a surge in fraudulent ads using images generated by artificial intelligence (AI) of celebrities to scam users. According to Monika Bickert, Meta’s VP of content policy, “This is a real-time process. It’s faster and more accurate than manual review.”

How It Works

Users locked out of their accounts can now upload a video selfie, which Meta will compare to their profile picture to verify their identity. Although this method isn’t a mandatory option, users can still recover their accounts by submitting official ID documents.

Additionally, the tool will be employed to detect and block scam ads that misuse celebrity images by comparing the faces in ads to public figures’ profiles. This is due to a rising phenomenon of ‘celebrity bait’ frauds that utilize the likeness of purposeful deepfakes or AI renderings of individuals. The new system will reportedly target these in real-time.

Early test results have been promising, with Meta already working with a small group of celebrities. The system automatically enrolls these figures, allowing them to opt out if desired.

Though effective, David Agranovich, Meta’s global director of security policy, admits it’s a game of numbers as scammers continuously adapt their methods. “Scam networks are highly motivated to just keep throwing things at the wall in hopes that things get through,” he said.

Regulatory Roadblocks and Privacy Concerns

Meta’s facial recognition rollout won’t be global—at least, not initially. Due to strict data protection laws and ongoing regulatory discussions, the technology will not be available in the European Union, the United Kingdom, or in U.S. states like Illinois and Texas, where biometric data laws are among the strictest. These legal hurdles stem from lawsuits Meta faced over its previous use of facial recognition, which resulted in settlements exceeding $2 billion.

Meta’s decision to shut down its long-running facial recognition program in 2021, which had been used to automatically tag users in photos, came amid growing public scrutiny over the technology’s potential misuse and privacy violations.

Agranovich emphasized that the company is now being careful with how it reintroduces the technology: “We don’t use the facial data for any purpose other than one-time scam verification processes. We also immediately delete any face data generated from ads, regardless of whether our systems find a match”.

The technology is expected to be deployed more widely in 2025, barring any further legal or regulatory challenges. In the meantime, Meta continues to refine its scam detection and account recovery processes to provide a more secure experience for its users.