Meta Deletes Face-Recognition System From Its Smart Glasses App (wired.com) 27
Last Thursday, Wired reported that Meta had quietly embedded an unreleased facial recognition system called NameTag into software installed on millions of phones. In a follow-up report, Wired says the tech giant has now removed the face-recognition-related code, while saying "no final decision" has been made about whether the feature will launch. From the report: On Thursday, WIRED reported that Meta had quietly integrated substantial portions of the NameTag system into the Meta AI app. Though never publicly enabled, the feature was designed to convert faces captured by the glasses into unique biometric signatures, commonly known as faceprints, and compare them against a database of faceprints stored on the user's device. WIRED also found that faces the system failed to recognize were cropped, indexed, and stored locally for future processing.
NameTag first surfaced in February, when The New York Times, citing internal Meta documents, reported that the company was developing face recognition for its smart glasses and weighing a launch as soon as this year. One memo reportedly described releasing it during a "dynamic political environment," when privacy and civil liberties advocates would be distracted. Last week, WIRED reported that much of NameTag's machinery was already built into the Meta AI app, downloaded by millions of users, as early as January, even as Meta publicly said it had made no final decision about face recognition. After WIRED's report, Stone dismissed the findings, writing that the company couldn't answer questions about how the system would work because "the feature does not exist." Andrew Bosworth, Meta's chief technology officer, called the reporting "incredibly misleading" and "absolutely dishonest."
[...] The newly released version of Meta AI removes nearly all traces of the feature Meta said did not yet exist. Gone is the face-recognition software itself, along with the code that ran the NameTag recognition process and the "Person recognized" alert the app would have shown if someone were identified. The update also strips out a folder where the app would have stored the cropped images and biometric signatures of faces it captured but could not identify. [...] A few fragments of the NameTag system remain in the version of latest Meta AI, including an internal debug menu label and a dormant link meant to open a recognized person's profile. The leftover code points to parts of the system that are no longer there.
NameTag first surfaced in February, when The New York Times, citing internal Meta documents, reported that the company was developing face recognition for its smart glasses and weighing a launch as soon as this year. One memo reportedly described releasing it during a "dynamic political environment," when privacy and civil liberties advocates would be distracted. Last week, WIRED reported that much of NameTag's machinery was already built into the Meta AI app, downloaded by millions of users, as early as January, even as Meta publicly said it had made no final decision about face recognition. After WIRED's report, Stone dismissed the findings, writing that the company couldn't answer questions about how the system would work because "the feature does not exist." Andrew Bosworth, Meta's chief technology officer, called the reporting "incredibly misleading" and "absolutely dishonest."
[...] The newly released version of Meta AI removes nearly all traces of the feature Meta said did not yet exist. Gone is the face-recognition software itself, along with the code that ran the NameTag recognition process and the "Person recognized" alert the app would have shown if someone were identified. The update also strips out a folder where the app would have stored the cropped images and biometric signatures of faces it captured but could not identify. [...] A few fragments of the NameTag system remain in the version of latest Meta AI, including an internal debug menu label and a dormant link meant to open a recognized person's profile. The leftover code points to parts of the system that are no longer there.
To Funny! (Score:3)
Re: (Score:1)
Easiest way to delete a feature (Score:5, Insightful)
Is to remove the user's access to it. We can still log the data into the cloud of course.
Re:Easiest way to delete a feature (Score:5, Insightful)
the problem is (Score:4, Interesting)
The problem is that this isn't very hard to do these days. It's pretty near impossible to prevent things that are easy to do.
The Meta device is constantly getting a stream of image frames from the camera in the glasses. Probably their device has enough compute horsepower to detect human faces, smartphones sure do. The faces can easily be cropped out of the images and passed along to whatever recognition system you happen to have on hand to develop a faceprint. It all goes into a database, local or remote, and then its a SMOP (simple matter of programming) to correlate a faceprint to a human identity. Gather all of that into a central database and presto.
You could just wander around with your cellphone in your shirt pocket recording everything and an app there could do much of this. Meta is getting some pushback because they are so visible and pervasive, but smaller players could definitely implement a mobile facial recognition system under the radar and probably have.
Re: (Score:3)
Google does it. Google photos identifies people and scans their face, puts it in your album. Matches them up for it's slide shows or other things.
It's not new tech. People just think they're safer because they can tell when someone raises a phone to record them, but completely forget about cameras designed to be discreet.
Re:the problem is (Score:4, Interesting)
You might be able to homebrew a device to recognize a face, but you won't be able to associate that with an identity, short of API access to Facebook or similar data brokers.
It sounds like you might be proposing manually assigning a name to each faceprint, but that's not what people are worried about. They're worried about random people who don't already know their name, getting it, and the associated data, instantaneously.
Re: (Score:3)
Someone has to initially assign a face to an identity, I agree. But once you've done that the pairing can be shared, either by you or by the app that you used to assign the ID, which in this case was Meta.
And then pretty soon you can be getting facial ID's on people you don't know but someone else did.
Re: (Score:2)
Someone has to initially assign a face to an identity
Easy to do automatically if you catch them exiting a car from the driver's seat. Lookup the car registration details. It's not perfect but it's good enough. Knowing what house they came out of narrows it down significantly too. You can find out basic details of everyone who lives in at a specific address. From that it's normally easy to guess who is who based on a random photo.
Re: (Score:2)
So you're proposing the construction of, maybe not an "open", but a distributed and crowdsourced faceprint database.
Unless you're restricting it to a small geographical area or niche, you would need a large number of people contributing to have any hope of random people entering your FOV being in the database. A large enough number, that many of them will be feeding junk data into the system. I would, for example.
Re: (Score:2)
I'm not 'proposing' a distributed and crowdsourced faceprint database. I'm merely pointing out that something like that could be done by a reasonably knowledgeable person using widely available tech. With AI assistance I could probably come up with a cellphone app coupled to a cloud server that does the basics. It's simple enough that it is not preventable.
You can argue that accurately assigning face prints to human identities could be spoofed without appropriate safeguards, but that's a different issue.
Re: the problem is (Score:1)
Perhaps Zuck-puppets?
Re: (Score:2)
The problem is that this isn't very hard to do these days. It's pretty near impossible to prevent things that are easy to do.
The Meta device is constantly getting a stream of image frames from the camera in the glasses. Probably their device has enough compute horsepower to detect human faces, smartphones sure do.
This wasn't very hard to do 20 years ago. The Viola-Jones [wikipedia.org] face detection (not recognition) algorithm was first described 25 years. I remember my pre-smartphone digit camera drawing boxes around the faces it could detect in a group photo, and letting me select who to focus on. Here's a 2007 article [adorama.com] about it, from one month before the iPhone's launch.
For now... (Score:4, Insightful)
Why would anyone think this "deletion" is permanent? As soon as the hubbub dies down, it'll be back.
The only way... (Score:2)
The only way I'm giving AI access to that level of personal information/interaction is if I own - and retain EXCLUSIVE access to the data.
Basically, the AI and data are mine and no one but me ever has access. No spying. No viewing. No data mining, anonymous or not. ZERO access during normal use unless I intentionally share something with a specific person or company. Think zero-knowledge encryption, but for my 'personal AI'.
Granted, companies want you to use AI largely so they can mine your data and it's g
Re: (Score:3)
The only way I'm giving AI access to that level of personal information/interaction is if I own - and retain EXCLUSIVE access to the data.
It's not the owner/wearer of the glasses that I'm worried about. It's all the poor schmucks they pass by (and photograph) on the street.
Re: (Score:2)
This. I guess Meta is finally getting smarter about catching bad publicity ahead of time. Say your a bad actor and load in every manager of, say a well known (evil) medical insurance company. How about looking up the Linkedin profile of the person that personally denied a family members cancer claim. Hell, load up evey person you have ever wanted to punch in the face.
To be honest though, I could really use this feature. I am constantly forgetting peoples names and other information. It be nice to hav
How ... (Score:2)
The newly released version of Meta AI removes nearly all traces of the feature Meta said did not yet exist.
... do you remove something that does not exist?
Yesterday, upon the stair,
I met a man who wasn't there
He wasn't there again today
I think he's with the CIA...
It is clearly a premium feature (Score:2)
This is not for you peasants. Feasibility has been demonstrated. They just need to make sure the people who still have it are not distinguishable from the masses who don't.
Re: (Score:2)
Certainly the Secret Service would love to have an automatic head's-up on the true identity of anyone walking into the room. Not that I would expect this implementation to be reliable enough actually to be used that way, but someday one will be.
Facial recog (Score:3)
As someone who has truly incredible trouble recognising people, this is kinda sad. Iâ(TM)d love to have a gadget that told me âoethatâ(TM)s George Stevenson, he lives just down the road and has a sick daughterâ so I donâ(TM)t look like a total prat.
I am not entirely face blind, though some unfortunates certainly are, but close enough to make life difficult.
So yeah, I was actually looking forward to precisely that feature.
Re: (Score:2)
Think of the counter example. Someone uses this to lookup a bunch of info on you then pretends you're old friends who bumped into each other. It'll make the initial kidnapping of or the ability to scam people much easier. Though it'll also making tracking down missing people easier... Everything has it's pros and cons.
"Nearly" (Score:2)
"Nearly" - "apart from the bits we didn't, which we can then slip back in when we think no one is looking, or will care."
Who trusts a single thing these fuckers say or do any more? Personally, I trust them as about as far as I can throw a wet mattress up a spiral staircase.
They can take their wanky nonce goggles and shove them so far up their collective arse that they're tickling their tonsils. From be
So I can't recognize child's friend's mom (Score:2)
Without pulling my phone out of the pocket and taking a picture with software that a coding agent wrote for me that does exactly the same thing on device. Whom does this benefit? Government doesn't care about social graces and they can also pay 100x more for custom built surveillance glasses. All Meta is doing in blocking me from avoid constant social embarrassment from difficulty placing faces.
\o/ (Score:1)
So owners of these glasses get no creepy benefit - they act as image harvesters for meta who run the face detection on their servers and share the live searchable database with 'law enforcement' etc?
I feel safer already.
> Find everyone wearing a Guy Fawkes mask in Central London right now ... Searching