Meta has announced it is trialling the use of facial recognition technology to prevent scams and allow users to recover their accounts more easily.

The parent company of Facebook and Instagram said that scammers often try to use images of public figures such as celebrities or influencers to trick people into visiting scam websites where they share personal information or send money.

While the social media giant already uses machine learning and automated technology to review the ads on its platforms, it said that this “celeb-bait” can be hard to detect as they are often very realistic.

To combat this scam, Meta said that it will use facial recognition technology on suspected scam ads to compare faces in the advert to the public figure's Facebook and Instagram profile pictures. If the technology confirms a match and determines the ad is a scam, Meta will block it.

The company added that it will immediately delete any facial data generated from ads for the one-time comparison, regardless of whether its system finds a match. The data will not be used for any other purpose.

Elsewhere, Meta said that people can lose access to their social media accounts if they forget their password, lose their device or are tricked into giving their password over to a scammer.

Should Meta believe an account has been compromised, it currently requires the account older to verify their identity before regaining access by uploading an official ID or an official certificate that includes their name.

The company is now testing video selfies as a way for people to verify their identity and regain access to their compromised accounts.

Meta will ask users to upload a video selfie which is then analysed by facial recognition technology to compare the image to the profile pictures on the account they’re trying to access in a similar way to how people use the tech to unlock phones or other apps.

It said that while hackers will keep trying to exploit account recovery tools, the new verification method will be harder for hackers to abuse than traditional document-based identity verification.

---