In partnership with
Some teams never seem to stop moving. They're on Attio, the agentic CRM.
Every customer signal is captured in one shared context layer, always current and compounding. Agents and workflows build pipeline, chase every buying signal, and move deals forward, an always-on revenue engine running alongside your team.
With Attio, you’ll get:
Leads automatically prioritised and routed to the right rep
Expansion and risk signals caught the moment they land
Follow-ups written in your voice, already there when you arrive
Teams like Parallel, Turbopuffer, and Wordsmith build on Attio. Are you one of them?
AI & Technology Intelligence
AI NEWS INSIDER
Issue #74 · September 14, 2026
THIS WEEK: A new lawsuit alleges Meta quietly built facial recognition technology from millions of photos, including people who never had an account. Here's the five-minute checklist worth running today.
6 MIN READ | Sharp, actionable intelligence
Meta Secretly Built a Face-Recognition Tool From Your Photos. Here's What to Check Today.
A new lawsuit alleges Meta used millions of Facebook and Instagram photos, including of people who never signed up, to build facial recognition technology for its smart glasses. The feature never shipped, but the data collection allegedly already happened.
By AI News Insider Editorial · 6 min read
A new lawsuit filed this month accuses Meta of doing something most of its users never agreed to: quietly extracting faceprints from millions of Facebook and Instagram photos to build a facial recognition system, without telling the people in those photos, including people who were never Meta users at all.
What NameTag actually is
The technology at the center of the case is called NameTag, designed to work with Meta's Ray-Ban and Oakley smart glasses. In theory, it would let someone wearing the glasses point them at a stranger and have Meta's AI identify who that person is. Code for the feature was discovered sitting inactive inside Meta's companion app in June, and Meta pulled it within a day of the discovery being reported. Meta says NameTag has never shipped to consumers and that no final decision has been made on releasing it.
The part that already happened, according to the lawsuit
The lawsuit doesn't hinge on whether NameTag ever ships. It alleges Meta already extracted biometric identifiers, essentially digital faceprints, from user photos to build and test the system, alongside training its Emu and Muse Image generative models on the same images. Critically, the 66-page complaint covers more than just Meta's own users. It includes anyone whose face appeared in a photo uploaded by someone else, meaning you don't need a Meta account to be named in this case. The proposed class covers photos uploaded since September 2021.
Why this keeps happening to Meta specifically
Illinois has one of the strictest biometric privacy laws in the country, the Biometric Information Privacy Act, which requires explicit written consent before a company collects or uses someone's faceprint, and lets individuals sue directly over violations. Meta already paid Texas $1.4 billion in 2024 to settle a similar claim over an earlier facial recognition feature. The pattern suggests biometric data collection remains an area Meta treats as worth the legal risk, until a court says otherwise.
What you can actually do about it today
You can't undo photos already uploaded, and neither this lawsuit's outcome nor Meta's public statements change what may have already happened to images already on the platforms. But there are concrete settings worth checking regardless of how the case resolves, covered in the checklist below.
The Bottom Line for AI News Insider Readers
This case may take years to resolve, and Meta disputes the allegations. But the underlying practice, training AI systems on user photos without clear, specific consent, isn't unique to one company or one lawsuit. The F.A.C.E. checklist below takes about five minutes and applies whether or not you've ever heard of NameTag.
You don't need to wait for a court ruling to reduce your own exposure. Four checks, about five minutes total.
F. Find and disable Facebook's face recognition setting
Facebook has had a face recognition on/off toggle in settings since an earlier privacy settlement. If you've never checked it, look under Settings & Privacy > Settings > Face Recognition, and turn it off if you'd rather Meta not use this technology on your account.
A. Audit what Meta AI is allowed to train on
Meta provides a form, and automatic opt-outs in some regions like the EU, letting you object to your data being used to train generative AI models. Check your regional settings; the process is more involved outside GDPR-covered regions, but it exists.
C. Check your Download Your Information archive
Meta's data download tool lets you see categories of information stored about you. It won't necessarily show a "faceprint" line item, but reviewing what's collected is a useful baseline before deciding what to do next.
E. Extend the same caution to people who never opted in
The lawsuit's most striking detail is that non-users, people who never made a Meta account, are part of the proposed class simply because their faces appeared in someone else's uploaded photo. Before you tag or upload a photo of a friend, a colleague, or your kids, remember that your own privacy settings don't cover people who aren't you.
Your Monday Morning Action
Open your Facebook settings right now, search "face recognition," and check what's currently turned on for your account. It's a two-minute check regardless of how this lawsuit turns out.
Business
Anthropic Tells Investors It Expects to Turn Profitable by Q2 2027
The company reportedly cited gross margins above 80% and roughly $11.5 billion in revenue this year, a sharp contrast to the burn-rate story dominating OpenAI's own path toward a public listing.
Security
AI Agents Used to Breach 395 Companies Across 48 Countries in Days
Researchers say a threat actor used AI agents built on OpenAI's Codex and a DeepSeek model to autonomously exploit a vulnerability, reaching full domain access at some targets in under two hours and compromising 11 organizations in a single 26-second burst at peak speed.
Safety
Anthropic and OpenAI Call for a Slowdown in AI Development
Both companies say they're introducing measures to better understand AI risk following a string of recent security incidents. Asian stocks and Nasdaq futures dipped in response, reflecting how closely markets are now tracking AI safety signals.
Big Tech
Amazon Opens a New AI Research Lab in San Francisco
The Amazon AGI SF Lab will focus on building foundational capabilities for AI agents, another sign of the biggest tech companies racing to build agent infrastructure rather than just chat interfaces.
66
Pages in the Class-Action Complaint Against Meta
Covering the alleged NameTag facial recognition system and AI training on user photos.
5 yrs
Span of Photos Covered by the Proposed Class
The complaint covers images uploaded since September 2021, including photos of non-users.
395
Organizations Breached by AI-Agent-Driven Attacks in Days
Across 48 countries, per GreyNoise researchers tracking the recent Codex-based exploitation campaign.
80%+
Gross Margins Anthropic Reportedly Told Investors to Expect
En route to profitability by Q2 2027, a notably different financial story than OpenAI's.
DROP (California's Delete Request and Opt-Out Platform)
One request, every registered data broker, no per-company forms
What it is: A free platform run by the California Privacy Protection Agency at cppa.ca.gov that lets California residents submit a single deletion request reaching every registered data broker at once, rather than contacting each one individually.
Why it matters now: Registered brokers were required to start processing DROP deletion requests as of August 1, 2026. It won't touch data Meta or another platform holds directly on you, but it does reach the wider data-broker ecosystem that scrapes and resells profiles, including biometric identifiers, built from scraped or purchased information.
Who gets it: Any California resident who wants to reduce their footprint across data brokers in one pass, rather than hunting down opt-out forms one company at a time.
• Single request reaches every registered California data broker at once
• Free to use, run directly by the state privacy regulator
• Pair it with Global Privacy Control, a browser signal that auto-communicates opt-out and is legally binding in several states
"People shouldn't have to worry if their biometric information will be misused."
Justin Boley · Partner, Wexler Boley & Elgersma, attorney for the plaintiffs in Alvarez v. Meta · 2026
Share AI News Insider
Know someone who uploads photos of other people without a second thought? Forward them this issue.
SHARE & EARN REWARDS →AI News Insider · Your weekly edge in artificial intelligence
aiinsider247.com
/