EU AI Act Compliance — Article 50 & Annex IIIThis AI Transparency Notice is published in accordance with the EU Artificial Intelligence Act (Regulation 2024/1689), which entered into force on 1 August 2024. Obligations under Article 50 (transparency for certain AI systems) apply from 2 August 2026. This notice describes how Investra uses AI, which systems are deployed, what rights users have, and how to exercise human oversight.
Version 4.0 — Effective July 24, 2026Updated to include full AI system register (6 systems), human oversight procedures, AI-generated content labelling, opt-out mechanisms per system, third-party AI provider disclosure, prohibited AI use policy, and EU AI Act compliance timeline.
1. Scope & Legal Basis
This notice applies to all AI systems deployed by Investra International Ltd (UK) and its Master Franchise partners across all Investra platforms. The legal bases for AI processing are:
| Legal Basis | GDPR Article | Application |
|---|
| Legitimate interest | Art. 6(1)(f) | Property recommendations, fraud detection, search ranking |
| Contract performance | Art. 6(1)(b) | Booking automation, KYC verification |
| Explicit consent | Art. 6(1)(a) + Art. 9(2)(a) | Personalised marketing, behavioural profiling |
| Legal obligation | Art. 6(1)(c) | AML/KYC automated screening |
2. EU AI Act Compliance Timeline
| Date | Obligation | Investra Status |
|---|
| 1 Aug 2024 | AI Act enters into force | ✅ Monitoring commenced |
| 2 Feb 2025 | Prohibited AI practices banned (Art. 5) | ✅ Compliant — no prohibited systems deployed |
| 2 Aug 2025 | GPAI model obligations (Art. 51–56) | ✅ No proprietary GPAI models — third-party only |
| 2 Aug 2026 | High-risk AI obligations (Art. 50 transparency) | 🟡 In progress — this notice published |
| 2 Aug 2027 | Full high-risk AI system compliance | 📅 Planned |
3. AI System Register
Investra currently deploys the following AI systems:
System 1 — Property Recommendation Engine
| Attribute | Details |
|---|
| Purpose | Recommends properties to users based on search history, saved properties, and behavioural signals |
| AI Type | Collaborative filtering + content-based recommendation |
| EU AI Act Risk | 🟢 Minimal risk |
| Data Used | Search queries, property views, saved listings, location preferences, budget range |
| Human Oversight | Editorial team reviews recommendation quality monthly |
| Opt-Out | Available via /account/privacy — disables personalised recommendations |
| Provider | Investra internal system |
System 2 — AI Property Valuation Assistant
| Attribute | Details |
|---|
| Purpose | Provides indicative property valuations based on comparable sales, location data, and market trends |
| AI Type | Regression model + comparable sales analysis |
| EU AI Act Risk | 🟡 Limited risk — output is indicative only, not binding |
| Data Used | Property characteristics, location, comparable sales, market indices |
| Human Oversight | All valuations marked as "AI-estimated — not a formal valuation". Users advised to obtain RICS valuation for transactions. |
| Opt-Out | N/A — informational only, no personal data profiling |
| Disclaimer | AI valuations are not formal appraisals and should not be used as the sole basis for investment decisions |
System 3 — AI Content Generation (Articles & Descriptions)
| Attribute | Details |
|---|
| Purpose | Generates property descriptions, market reports, and editorial content |
| AI Type | Large Language Model (LLM) |
| EU AI Act Risk | 🟢 Minimal risk — content is labelled |
| Labelling | All AI-generated content is marked with "AI-assisted" label per EU AI Act Art. 50(4) |
| Human Oversight | Editorial review required before publication of all AI-generated articles |
| Third-Party Provider | OpenAI (GPT-4), Google (Gemini) — see Section 5 |
System 4 — KYC/AML Automated Screening
| Attribute | Details |
|---|
| Purpose | Automated screening of users against sanctions lists, PEP databases, and adverse media |
| AI Type | Rule-based + ML classification |
| EU AI Act Risk | 🔴 High risk — affects access to services |
| Human Oversight | All positive matches reviewed by compliance officer within 24 hours before action taken |
| Right to Contest | Users may contest automated screening results — contact compliance@investra.io |
| Legal Basis | Legal obligation (AML Directive 2015/849 as amended) |
| Third-Party Provider | Refinitiv World-Check / ComplyAdvantage |
System 5 — Fraud Detection
| Attribute | Details |
|---|
| Purpose | Detects fraudulent listings, fake accounts, and suspicious transaction patterns |
| AI Type | Anomaly detection + behavioural analysis |
| EU AI Act Risk | 🟡 Limited risk |
| Human Oversight | Automated flags reviewed by trust & safety team within 48 hours |
| Right to Contest | Users whose accounts are flagged may appeal — contact trust@investra.io |
System 6 — AI Chatbot / Virtual Assistant
| Attribute | Details |
|---|
| Purpose | Answers user queries about properties, platform features, and booking processes |
| AI Type | LLM-based conversational AI |
| EU AI Act Risk | 🟢 Minimal risk |
| Disclosure | Chatbot identifies itself as AI at the start of every conversation per EU AI Act Art. 50(1) |
| Human Escalation | Users can request human agent at any time — response within 24 hours |
| Third-Party Provider | OpenAI (GPT-4 API) |
4. AI-Generated Content Labelling
In compliance with EU AI Act Article 50(4), all AI-generated or AI-assisted content on Investra platforms is labelled as follows:
| Content Type | Label | Placement |
|---|
| AI-generated articles | "AI-assisted content" | Byline / article header |
| AI property descriptions | "Description generated with AI assistance" | Below property description |
| AI valuations | "AI-estimated value — not a formal valuation" | Adjacent to valuation figure |
| AI chatbot responses | "You are chatting with an AI assistant" | Chat window header |
| AI-generated images | "AI-generated image" | Image caption |
5. Third-Party AI Providers
| Provider | System | Data Shared | Location | Safeguards |
|---|
| OpenAI (GPT-4) | Content generation, chatbot | Anonymised text prompts | USA | SCCs + OpenAI DPA |
| Google (Gemini) | Content generation | Anonymised text prompts | EU/USA | SCCs + Google DPA |
| Refinitiv / ComplyAdvantage | KYC/AML screening | Name, DOB, nationality | EU/USA | SCCs + DPA + GDPR Art. 28 |
| Stripe Radar | Fraud detection (payments) | Transaction data | EU/USA | SCCs + Stripe DPA |
All third-party AI providers are bound by Data Processing Agreements (DPAs) and Standard Contractual Clauses (SCCs) where data is transferred outside the EEA.
6. Profiling & Automated Decision-Making
Investra uses automated processing that may produce legal or similarly significant effects in the following cases:
| Use Case | Automated Decision? | Human Review? | Right to Contest |
|---|
| KYC/AML screening — match found | Yes — account suspended pending review | Yes — within 24h | compliance@investra.io |
| Fraud detection — account flagged | Yes — access restricted pending review | Yes — within 48h | trust@investra.io |
| Property recommendations | Yes — personalised ranking | No — informational only | Opt-out via /account/privacy |
| Search ranking | Yes — relevance scoring | No — informational only | Opt-out via /account/privacy |
For decisions with significant effects (account suspension, access restriction), Investra guarantees human review before any final decision is implemented. Users have the right under GDPR Art. 22 to not be subject to solely automated decisions that produce legal effects.
7. Human Oversight Procedures
Investra maintains the following human oversight procedures for all AI systems:
- KYC/AML: Compliance officer reviews all positive matches within 24 hours. No account action taken based solely on automated screening.
- Fraud detection: Trust & safety team reviews all flags within 48 hours. Users notified of any account restrictions.
- Content generation: Editorial team reviews all AI-generated articles before publication. Factual claims verified against primary sources.
- AI valuation: Valuations are marked as indicative. Users advised to obtain independent professional valuation for transactions.
- Chatbot: Human escalation available at any time. Complex queries routed to human agents.
- Monthly review: AI system performance reviewed monthly by Investra's AI governance committee.
8. User Rights & Opt-Out Mechanisms
| Right | Legal Basis | How to Exercise |
|---|
| Opt out of personalised recommendations | GDPR Art. 21 (legitimate interest) | /account/privacy → disable Analytics & Marketing |
| Opt out of behavioural profiling | GDPR Art. 21 | /account/privacy → disable Marketing cookies |
| Contest automated KYC/AML decision | GDPR Art. 22 | compliance@investra.io — response within 24h |
| Contest fraud detection flag | GDPR Art. 22 | trust@investra.io — response within 48h |
| Request human review of any automated decision | GDPR Art. 22(3) | dpo@investra.io |
| Access AI processing records about you | GDPR Art. 15 | dpo@investra.io — DSAR form |
| Request explanation of AI decision | EU AI Act Art. 86 | dpo@investra.io |
9. Prohibited AI Uses
In compliance with EU AI Act Article 5, Investra does not deploy and will never deploy:
- AI systems that use subliminal techniques to manipulate user behaviour
- AI systems that exploit vulnerabilities of specific groups (age, disability, social situation)
- Social scoring systems that evaluate users based on social behaviour or personal characteristics
- Real-time remote biometric identification systems in publicly accessible spaces
- AI systems for emotion recognition in employment or education contexts
- AI systems that create or expand facial recognition databases through untargeted scraping
- Predictive policing systems based solely on profiling
10. AI Risk Classification
| System | EU AI Act Risk Level | Rationale |
|---|
| Property Recommendation Engine | 🟢 Minimal | Informational — no significant effects on users |
| AI Valuation Assistant | 🟡 Limited | Indicative only — no binding decisions |
| AI Content Generation | 🟢 Minimal | Labelled — human review before publication |
| KYC/AML Screening | 🔴 High | Affects access to services — human oversight mandatory |
| Fraud Detection | 🟡 Limited | Flags for human review — no automatic permanent action |
| AI Chatbot | 🟢 Minimal | Disclosed as AI — human escalation available |
11. Bias Monitoring & Fairness
Investra is committed to fair and non-discriminatory AI systems. Our bias monitoring programme includes:
- Monthly audits: Recommendation engine audited for demographic bias (nationality, language, budget)
- KYC/AML fairness: False positive rates monitored by nationality and ethnicity to detect discriminatory patterns
Fraud detection:Monitored for disproportionate impact on specific user groups- External audit: Annual independent bias audit by qualified third party
- Reporting: Bias audit results published in annual transparency report
If you believe an AI system has treated you unfairly or discriminatorily, contact dpo@investra.io.
12. AI Incident Reporting
Investra maintains an AI incident log and will report serious AI incidents to the relevant national supervisory authority as required by EU AI Act Article 73. Users may report AI incidents to:
- Email:ai-safety@investra.io
- Response time: Acknowledgement within 24 hours, investigation within 7 days
- Regulatory reporting: Serious incidents reported to ICO (UK) and relevant EU national authority within 15 days
13. Contact & DPO
Change Log
| Version | Date | Changes |
|---|
| v4.0 | July 24, 2026 | Full AI system register (6 systems), EU AI Act compliance timeline, third-party AI providers table, prohibited AI uses (Art. 5), risk classification table, bias monitoring programme, AI incident reporting, human oversight procedures per system, opt-out table with direct links |
| v3.0 | July 24, 2026 | Initial version with EU AI Act Art. 50 compliance |