Password free. Next level security with biometric intelligence
We are developing Avatar Me to bring biometric intelligence into a secure, continuous relationship between people and the systems acting on their behalf. At FT1 Labs, this work is part of our ambition for the advanced INTELLICA OS Platform: connected identity, clear permissions and human control.
A password asks a person to remember a secret. An intelligent system acting on someone’s behalf needs a more demanding answer: who is present, what have they authorised, and is that authority still valid? Avatar Me is our research direction for that relationship.
A shared secret has limits
Passwords can be disclosed, reused and entered into fraudulent websites. Passwordless authentication already exists: correctly implemented cryptographic passkeys can resist phishing without sending a reusable password to a service. Biometrics can help unlock a device-held credential; they are not themselves a secret or a replacement for the underlying security protocol.
NIST’s digital identity guidance describes these distinctions and sets limits on biometric use. A recognisable face or voice is evidence to assess, not proof that a person has approved an action.
Source: NIST Digital Identity Guidelines ↗Recognition is only one part of identity
Face, iris and voice signals describe different aspects of a person. Presence checks and behavioural signals may add context, but context can be incomplete. A familiar voice in a recording is not the same as a person participating in a live interaction.
Combining signals may help a system assess uncertainty, yet more inputs do not automatically make authentication safe. Sensors can fail together; lighting, illness, noise, disability and changes in appearance can affect the experience. The useful question is which evidence is appropriate for the action, and how a person can proceed when it is unavailable.
“Being recognised should never mean surrendering control.”
A personal avatar, with boundaries
Avatar Me explores a personal digital identity that could connect a person to the intelligent systems they choose to use. The avatar is a way to express permissions and continuity; it is not a public replica or permission to impersonate its owner.
Our concept brings multiple modes of recognition together with device-bound credentials, explicit consent and a clear record of authority. A person would decide which services may participate, what each may do and when that access ends. Pausing recognition or revoking a permission should be as understandable as granting it.
These are design and research aims. Avatar Me is in development, with no claim here of a finished authentication product, validated security performance or commercial availability.
Privacy begins before the first scan
The UK Information Commissioner’s Office explains that biometric recognition used to uniquely identify a person processes special category biometric data. Organisations need both a lawful basis and an applicable special category condition. Consent is not a substitute for considering necessity, proportionality and the surrounding circumstances.
Source: ICO guidance on lawful biometric processing ↗Verify what is needed. Reveal less.
We are exploring verification that could return a narrowly scoped confirmation rather than distributing raw biometric recordings. Local processing, limited retention and separation between identity and activity are directions to investigate, not promises that every future deployment will operate in the same way.
People need to understand what is collected, where it is processed and how to stop. A practical system must also offer an accessible alternative to biometric recognition and a recovery route when a device is lost or a signal no longer works. This website does not capture biometric samples.
A voice can request an action. Authority must follow.
Voice-to-action intelligence creates a new boundary between conversation and consequence. Asking an assistant to prepare a booking is different from authorising a payment or unlocking a door. Recognising the speaker should not collapse those distinctions.
In our proposed direction, identity evidence would be considered alongside the requested action, the device and the person’s permissions. Consequential actions would need a clear confirmation tied to the exact transaction. The system should stop or ask when authority is uncertain, even when the voice sounds familiar.
An identity layer for the INTELLICA OS Platform
We are exploring Avatar Me as an integrated identity and authorisation layer for our advanced INTELLICA OS Platform. The aim is to connect biometric intelligence with a voice-first platform that understands intent, maintains context and coordinates tools and real-world actions.
In this research direction, Avatar Me would help establish who is present and which permissions apply, while INTELLICA interprets the request and prepares the appropriate action. A recognised identity would not confer unlimited authority: access would remain specific to the service, device and task, with additional confirmation for consequential decisions.
For example, a person could ask INTELLICA to prepare a journey, then review and authorise a particular booking through their chosen identity checks. The platform would need to keep an understandable record of that permission and make it possible to pause the connection or revoke access. Integration, security testing and recovery behaviour remain areas of research and development, rather than claims of a finished commercial capability.
Trust has to survive impersonation
Synthetic speech, manipulated video and replayed recordings make visual or acoustic familiarity an unsafe foundation for trust. A useful evaluation would examine presentation attacks, compromised devices, stolen credentials and social engineering together.
No single liveness check should be presented as an answer to every attack. Independent testing, measured false acceptance and rejection rates, accessibility assessment and secure recovery would be necessary before making performance claims. An identity system also needs a way to explain a refusal without revealing information that helps an attacker.
What changes next?
Imagine arriving in a familiar space and choosing to connect your private assistant. The environment recognises an authorised relationship, reveals the permissions in use and asks before doing anything consequential. Leaving ends the session. Losing a device does not mean losing the ability to recover your identity.
That is the experience Avatar Me sets out to explore. Our next questions concern the minimum evidence needed for each action, robust alternatives when recognition fails, and how people can inspect and revoke authority. Removing a login screen is only useful if people can still understand and control who acts on their behalf.
