
Facia
Facia is a UK facial biometrics platform combining liveness detection, deepfake detection and face recognition in a single API for businesses and governments. It runs in the cloud or on-premises and bills from $0.10 per check.
What is Facia?
Facia is a British facial biometrics platform that sells three closely related capabilities to businesses and governments: liveness detection, deepfake detection and face recognition. It is not a consumer product. Everything is consumed through a REST API, a web SDK or mobile SDKs, and it can run either in Facia's cloud or entirely on the customer's own servers.
The liveness engine answers one question: is the face in front of the camera a living person, right now? A single passive frame is enough, with no instruction given to the user, and the company reports an API response under one second and an end-to-end check of one to three seconds. Its DeepLiveness layer folds deepfake detection into that same check, combining RGB texture analysis, motion analysis, depth mapping and neural anomaly detection. Facia publishes a false acceptance rate of 0.06% and a false rejection rate of 0.3% after its latest algorithm update, claims cover against 56 or more spoofing attack types, from paper masks and silicone masks to wax figures, replay and injection attacks, and points to two independent references anyone can open: an iBeta Level 2 certification against ISO 30107-3 and a NIST FRTE report card.
The deepfake product works independently of identity checks. It scores images, recorded video and live video calls, and Facia reports roughly 90% average accuracy across the datasets it tested, against the 82% reached by the best model on Meta's DFDC benchmark. Media can be uploaded to Facia's own portal or scanned automatically through a backend integration, and a public Discord bot lets anyone try it.
The recognition side covers 1:1 photo-to-ID matching, 1:N face search, image-to-video and video-to-video matching, age estimation and document verification. Published use cases run from KYC onboarding and account de-duplication to access control, attendance, surveillance, immigration lanes, ticketing, pay-with-face and passwordless sign-in, with dedicated pages for banking, iGaming, dating apps, retail, government and event management. Founded in 2022 and based in London, Facia says it serves clients across 190 to 230 countries and territories, on models trained for demographic and racial diversity.
What it does
- Confirm that a face in front of a camera belongs to a live person, actively or passively
- Detect deepfakes and AI-generated faces in images, recorded video and live video calls
- Match a selfie against an identity document photo (1:1)
- Search a face across a large database (1:N) and eliminate duplicate accounts
- Estimate a person's age from facial features, without storing an identity document
- Verify identity documents as part of a KYC flow
- Trigger step-up face verification before a high-risk account action
When to use Facia / When not to
A quick filter to help you decide if Facia is the right fit.
When to use Facia
- KYC and onboarding teams in banking, fintech and iGaming that need a liveness check hardened against AI-generated faces
- Security and fraud teams fighting account takeover, synthetic identities and duplicate accounts through 1:N face search
- Trust and safety, newsroom and content moderation teams that must judge whether an image, a video or a live call is genuine
- Regulated or sovereignty-conscious organisations that need biometrics running on their own servers behind their own firewall
- Engineering teams integrating identity checks into an existing product through a REST API or lightweight mobile and web SDKs
When not to use Facia
- Individuals looking for a consumer app: access goes through a signup portal and, in practice, a sales conversation
- Small projects hoping for a permanent free tier, since the only free entry point is a 10-credit trial
- Buyers who need a published price above 1,000 checks per month, as enterprise volume pricing is quote-only
- Organisations that must contractually forbid vendor training on their data, which the terms of service make a condition of use
- Teams with a hard data-residency requirement, because no hosting country or region is named anywhere on the site
How to use Facia
A typical end-to-end flow, from setup to results.
- Decide which service you need: liveness, deepfake detection, face match, 1:N search, age estimation or document verification
- Open an account on portal.facia.ai and claim the 10 free credits, which require no credit card
- For API access, contact support@facia.ai, then use the password link sent back to you
- Call /request-access-token with your email and password, or your client ID and client secret, to obtain an access token
- Point your integration at the base URL https://api.facia.ai/, over HTTPS only, with JSON request and response headers
- Choose your delivery channel: REST API, the web SDK iframe, or a mobile SDK for Android, iOS, Flutter, React Native or Cordova
- Check the published browser compatibility matrix before shipping, since Firefox is unsupported and Opera only partially
- Watch the rate limits: 20 verifications per minute on Pay As You Go, 60 on Enterprise
- For a local deployment, request access through the on-premises form and work with the team on installation
- Move past 1,000 checks per month by requesting an Enterprise quotation, and monitor the service on status.facia.ai
Pros & Cons
Pros
- Independent certifications you can verify yourself: iBeta Level 2 against ISO 30107-3 and a public NIST FRTE report card
- Unusually transparent performance figures, including a 0.06% false acceptance rate and a 0.3% false rejection rate
- Per-check pricing published service by service, which is rare in enterprise biometrics
- A trial of 10 credits that needs no credit card
- Broad functional coverage in one contract: liveness, deepfake, face match, 1:N search, age and documents
- On-premises deployment available for organisations that cannot send faces to a vendor cloud
- Explicit attention to demographic and racial bias, with accuracy tuned for APAC, MENA and South Asian faces
Cons
- Using the service is treated as consent to have your data train Facia's AI models, with no documented opt-out
- No hosting country or region is ever named, only servers in multiple locations
- No public sub-processor list, even though sub-processing is contractually permitted
- No Article 27 EU representative is designated, despite European customers being targeted
- No permanent free plan, and no published pricing at all above 1,000 checks per month
- Deepfake accuracy figures differ between pages, quoted as 90%, 95% and 100% on DFDC depending on where you look
- Small credibility snags: the App Store link for the demo app is dead, and the California notice ships with blank contact fields
Pricing & Plans
There is no permanent free plan. New accounts receive 10 complimentary credits that require no credit card, which functions as a trial rather than a free tier. The lowest published price point is USD 0.10 per verification, applying to a Quick Liveness check or an age estimation under the Pay As You Go plan. Other unit rates run from USD 0.12 for a detailed liveness check to USD 0.70 for document verification. Above 1,000 checks per month, pricing moves to the Enterprise plan and is available on quotation only.
- Quick Liveness USD 0.10
- Detailed Liveness USD 0.12
- Age Estimation USD 0.10
- Face Match 1:1 USD 0.15
- Face Search 1:N USD 0.15
- Deepfake Detection USD 0.40
- Document Verification USD 0.70
- rate limited to 20 verifications per minute
- above 1
- 000 checks per month
- custom quotation announced as up to 10 times cheaper than individual rates depending on volume and requirements
- rate limited to 60 verifications per minute
Data, GDPR & hosting
A consolidated view of how Facia handles your data.
GDPR overview
GDPR handling is concrete but incomplete. Facia publishes a GDPR compliance certificate and a CCPA certificate as PDFs, states that it processes personal data under Article 28 of the EU and UK GDPR, and names a data protection officer reachable at dpo@facia.ai. The privacy policy lists the full set of rights, access, rectification, erasure, objection, portability, withdrawal of consent and restriction, with a one-month response target, and points complainants to the UK Information Commissioner's Office. Transfers outside the EEA are covered by standard contractual clauses, and Schedule 5 of the terms doubles as a data processing agreement. Two gaps matter: no Article 27 representative in the European Union is named anywhere, and no sub-processor list is published, despite sub-processing being contractually allowed.
Who owns the data?
The terms set up a classic processor arrangement: the client acts as controller, Facia as processor under Article 28 of the EU and UK GDPR, and Facia may only handle client data on documented instructions and for the purposes listed in Schedule 5 of its terms. The client warrants that it has the right to disclose the personal data it sends, and keeps ownership of its own brand assets, which Facia may nevertheless reuse for promotion. Facia can refuse an instruction it judges unlawful. What the documents never state in so many words is who owns the biometric data itself, so ownership is inferred from the processor role rather than declared.
Reuse rights
Reuse by the client is explicitly fenced in rather than free. Under the credential-use restrictions, verification data may only serve the specific purpose disclosed to the end user; it must not be kept longer than necessary, must not be combined with unrelated datasets, and must not be used to build an independent identity database without a lawful basis. Downstream use must comply with data protection law, and the client is responsible for notifying third parties and obtaining permissions, including for transfers outside the EEA. In the other direction, Facia grants itself a broad right: using the service is treated as explicit consent for Facia to use that data to optimise, improve and train its AI models, with anonymisation applied only where required. Facia may also engage sub-processors, subject to notification and a right to object.
Data retention & training
Hosting summary
Facia never names a hosting country or region. The terms of service say only that it maintains servers in multiple locations with automatic failover and a backup strategy, and that hosting and server infrastructure are delegated to an unnamed third-party provider whose downtime, outages and security failures Facia explicitly disclaims. The privacy policy adds that data may be transferred outside the United Kingdom, with a similar degree of protection promised, and the terms allow transfers beyond the European Economic Area under standard contractual clauses, with the client responsible for obtaining the necessary consents. On security, Schedule 5 commits to SSL and TLS in transit, AES 256-bit encryption or SHA-256 hashing at rest, and storage in repositories described as SSAE-compliant and ISO-accredited. The only configuration where the customer controls the jurisdiction is the on-premises deployment, where the software runs behind the customer's own firewall. Any buyer with a data-residency obligation should get the locations in writing before signing.
Things to keep in mind
Risks and trade-offs to weigh before adopting Facia.
- Your data trains their models: the terms make using the service equivalent to consenting to AI model training, with no documented way out
- Facial biometrics are sensitive personal data almost everywhere, so deploying Facia without a legal review is a genuine regulatory exposure
- You cannot tell where the faces end up: no hosting country is named, and infrastructure is delegated to an unnamed third party whose failures Facia disclaims
- Deletion is irreversible and self-service only by email, so an over-broad request can destroy records you still needed
- Accuracy figures that vary from page to page invite over-confidence; treat 0% false acceptance as a lab result, not a field guarantee
- Automated face decisions can fail unevenly across populations, and Facia itself concedes AI verification is not 100% error-free
- Building a 1:N face database is a surveillance capability first and a convenience second, and the terms put the lawful basis for it entirely on you
Setup & Integrations
Technical difficulty
Moderate, and developer work is unavoidable: there is no no-code path. Onboarding runs through an email to support, a password link, then a token from /request-access-token, after which you call api.facia.ai over HTTPS. Facia does the heavy lifting on the client side, with roughly 5 MB SDKs for Android, iOS, Flutter, React Native and Cordova that already handle camera and internet permissions, plus a web SDK delivered as an iframe. No special hardware is needed. Check the browser matrix first, as Firefox is unsupported. On-premises deployment needs infrastructure work but comes with 24/7 support.
Deployment
Apps stores
Integrations
Behind Facia
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Facia.
Frequently asked questions
What exactly does Facia do?
How much does Facia cost?
Is there a free trial or a free plan?
Which certifications does Facia hold?
How fast and how accurate is it?
Can Facia run on our own servers?
Which platforms and languages are supported?
Does Facia train its models on customer data?
How long is data kept, and can it be deleted?
Where is the data hosted?
Should you pick Facia?
Facia is one of the more credible players in a market crowded with unverifiable claims. Its two strongest arguments are not marketing lines but documents anyone can open: an iBeta Level 2 certification against ISO 30107-3 and a NIST FRTE report card hosted by NIST itself. Add published per-check pricing, which most enterprise biometrics vendors refuse to give, a 10-credit trial with no credit card, and an on-premises option for organisations that cannot send faces to a vendor cloud, and the commercial proposition is unusually legible.
The reservations are all about data governance rather than technology. The terms of service treat using the service as consent to have your data train Facia's AI models, and no opt-out is documented. No hosting country or region is named anywhere, no sub-processor list is published, and no Article 27 representative is designated in the European Union even though European customers are actively courted. For a vendor handling facial biometrics, a category that regulators treat as sensitive by default, those are not minor omissions.
A few smaller inconsistencies dent the polish: deepfake accuracy is quoted as 90%, 95% and 100% on DFDC depending on which page you read, the number of spoofing attacks covered is 53 in one place and 56 in another, the App Store link for the demo app is dead, and the California privacy notice ships with blank contact fields.
The sensible reading is that Facia deserves a serious evaluation for a B2B biometrics need, particularly where deepfake resistance matters, but that the data clauses should be negotiated explicitly rather than accepted as published. Ask for the hosting locations, the sub-processor list and a training exclusion in writing before signing anything.
- Choosing a selection results in a full page refresh.
- Opens in a new window.