Facebook is discussing building facial recognition into its upcoming smart glasses product and has been weighing the legal implications of the controversial technology, executives said at an internal meeting Thursday.

During a scheduled companywide meeting, Andrew Bosworth, Facebook’s vice president of augmented and virtual reality, told employees that the company is currently assessing whether or not it has the legal capacity to offer facial recognition on devices that are reportedly set to launch later this year. Nothing had been decided, he said, and he noted that current state laws may make it impossible for Facebook to offer people the ability to search for others based on pictures of their face.

“Face recognition … might be the thorniest issue, where the benefits are so clear, and the risks are so clear, and we don’t know where to balance those things,” Bosworth said in response to an employee question about whether people would be able to “mark their faces as unsearchable” when smart glasses become a prevalent technology. The unnamed worker specifically highlighted fears about the potential for “real-world harm,” including “stalkers.”

Following publication of this story, Bosworth wrote on Twitter that “we’ve been open about our efforts to build [augmented reality] glasses and are still in the early stages.”

“Face recognition is a hugely controversial topic and for good reason and I was speaking about was how we are going to have to have a very public discussion about the pros and cons,” he said.

Facebook has been touting a smart glasses product since last year, saying that it will arrive “sooner than later” in 2021. Built in partnership with Ray-Ban and its parent company, Luxottica Group, the wearable devices are expected to rival products from competitors Snapchat and Amazon, providing an augmented reality experience that will overlay digital features on people’s real-world perspectives.

Bosworth’s comments about a Facebook facial recognition device come at a time of heated debate and scrutiny of the technology, which has been deployed by federal agencies, police departments, and even private companies — often in the name of safety and security. Critics and academics have decried the lack of federal regulation over the technology, which they say could undermine individual privacy. Some cities, including San Francisco and Boston, have already banned it from government use.

Facebook, which has one of the largest repositories of user-uploaded photos, has taken a cautious approach to facial recognition. In 2015, it deployed a system called DeepFace that it’s since used to power features, such as photo tag suggestions, to identify people in pictures. Unlike Amazon, which sells its facial recognition tool to other companies and public entities, Facebook has not marketed DeepFace or any of its other facial recognition technologies for use beyond its own social network.

In his remarks Thursday, Bosworth said he has been discussing facial recognition with CEO Mark Zuckerberg and he was overseeing a privacy review for the smart glasses product later in the day. Late last year, the vice president also noted in an internal memo, first reported by the Big Technology newsletter, that the company should “differentiate our products on the basis of privacy.”

“We should become the undisputed leaders in providing privacy aware software,” Bosworth wrote in December.

While Bosworth said he understood privacy concerns over facial recognition on Thursday, he criticized current legislation, including Illinois’ Biometric Information Privacy Act (BIPA), which prevents private companies from collecting and storing biometric data like face scans without people’s consent. He noted the law was passed in 2008 and questioned how such out-of-date legislation could apply to the technologies of today.

Last year, Facebook was ordered to pay $650 million to Illinois citizens for violating BIPA with its photo-tagging feature.

“The real question is whether we will be able to recognize any faces at all, and we don’t know. Legally, the answer might be no, if you’re familiar with BIPA in Illinois … people are making face recognition illegal,” Bosworth told employees Thursday. “That’s ok. We can do that as a society and this product will survive and thrive without it. I do think there are some lost opportunities, though.”

In explaining possible use cases of facial recognition, Bosworth said that Facebook’s smart glasses may help someone recognize someone at a dinner party if they can’t remember their name or if they have face blindness.

During the conversation, Facebook Chief Diversity Officer Maxine Williams noted that Facebook may need to develop its own principles on facial recognition that would take precedence in places where there are currently no laws governing the technology. “Just because you can [build something,] doesn’t mean you will,” she added, noting that the company would have to consider the product’s potential for harm and discrimination.

As Facebook explores the possibility of adding facial recognition to future products, it’s been sparring with a company that scraped its photos and data to build its own facial recognition tool. Clearview AI, which has provided its software to hundreds of police departments, federal agencies, and private companies, trained its facial recognition tool, in part, on photos taken from Facebook and Instagram.

While Facebook said previously that it’s sent cease and desist letters to Clearview for violating its terms of service, the company has yet to take legal action against the New York City–based startup. Facebook board member Peter Thiel was an early investor in Clearview.