AI Governance Institute
← News

AI Companion Service Pairs Biometric Age Checks With Always-On Emotional Inference

What happened

Xicoia Ltd, the UK company behind a viral AI actress service called Tilly Norwood, requires callers to submit a video selfie processed by Spain-based Didit for automated age verification before any interaction begins, as reported by Bleaching Computer. The company relies on legitimate interests rather than explicit consent as its legal basis for this biometric processing under UK data protection law. During calls, the system infers users' emotional states through continuous camera and voice analysis, a feature the service does not allow users to disable. Xicoia frames this design as a response to the UK Online Safety Act's age-assurance requirements, which compel platforms likely accessed by minors to implement age verification. The arrangement also creates a cross-border biometric data flow from UK users to a Spanish processor, raising transfer-compliance questions that the service's public documentation does not fully address.

Why it matters

  • ·Using legitimate interests as the legal basis for face scanning before a call is contestable under UK GDPR. A supervisory authority challenge could invalidate the age-verification architecture and expose the operator to enforcement, regardless of whether the underlying age-check obligation is genuine.
  • ·Always-on emotional inference that cannot be disabled by users is a significant data-minimization and purpose-limitation problem. Under most data protection frameworks, including UK GDPR, affective computing outputs qualify as inferred sensitive data, and processing them without an opt-out mechanism is difficult to defend.
  • ·Enterprises procuring or deploying AI companion or persona services face a vendor due-diligence gap: this case shows that age-verification compliance in one jurisdiction can embed biometric and affective data practices that create independent liability in others, making third-party intake assessments more complex.

Stay ahead of stories like this

Get every UK AI governance development like this one, plus the rest of the week's developments. Every Thursday.

Free every Thursday. Unsubscribe anytime.

Governance controls affected

What to do now

  • ☐Audit any AI companion or persona services in your vendor portfolio to determine whether age verification is implemented using biometric processing and what legal basis is documented.
  • ☐Review whether any procured or deployed AI services perform continuous affective inference and confirm that users have a meaningful opt-out mechanism; flag services that do not for escalation.
  • ☐Assess cross-border data flows where biometric data collected in one jurisdiction is processed by a third-party vendor in another, and verify that transfer mechanisms are documented and adequate.
  • ☐Update your third-party AI vendor intake questionnaire to explicitly ask about age-verification methods, the legal basis for biometric processing, and whether emotional inference features can be disabled.
  • ☐Brief your data protection officer on the legitimate-interests-as-basis-for-face-scanning pattern so your organization does not replicate this design in any internally deployed AI service.

What to watch next

The UK Information Commissioner's Office has not yet issued definitive guidance on whether legitimate interests is a permissible basis for biometric age verification under Online Safety Act-mandated schemes, and enforcement posture on affective computing remains unsettled. Compliance teams should monitor ICO outputs under the Regulations Requiring the ICO to Produce a Statutory AI and Automated Decision-Making Code of Practice and watch whether Ofcom's age-assurance standards begin specifying permissible data-processing architectures. The Anthropic age assurance policy covered in Anthropic's Age Assurance Policy Shifts Minor-Access Compliance Onto Deployers is a related signal that minor-access obligations are being pushed down supply chains in ways that force deployers to make consequential data-architecture choices.

Related Coverage

Research2026-11-18

Fireworks AI Breach Exposes 149,000 Uber Drivers' License Data

Hackers used stolen credentials to access a dedicated cloud storage environment Uber had set up for AI inference vendor Fireworks AI, exposing personal data for approximately 149,346 Uber drivers. Stolen data included driver's license images, names, home addresses, dates of birth, and license numbers. This was the second security incident at Fireworks AI in October 2026. Uber did not begin notifying affected drivers until more than a month after learning of the breach.

Enforcement2026-11-17

IBM MCP Gateway Flaw Sends User Identity to Servers Configured to Block It

IBM disclosed CVE-2026-110278 in its ContextForge MCP Gateway, versions 1.0.0 through 1.1.0. A flaw causes the gateway to ignore per-server settings that control which upstream MCP servers receive user identity information, such as user IDs, email addresses, and group memberships. IBM recommends immediate upgrade to version 1.1.1, or globally disabling identity propagation as a temporary workaround.

Research2026-11-15

Edge0-145B Puts Frontier AI on a Laptop, Exposing Shadow AI and Data Governance Gaps

Edge0 has open-sourced Edge0-145B, a 145-billion-parameter AI model designed to run on a compatible personal laptop without cloud infrastructure. The model requires a reported peak of 29GB of memory and is claimed to match the coding and reasoning performance of specific frontier cloud models. Because inference runs locally, no data is sent to a cloud provider, but governance teams also lose the audit trail and access controls that cloud deployments typically provide.