Information about AI

Last updated: 29 April 2026
Version: 2.1 Effective date: 29 April 2026

This Information about AI explains how GET HAUZED, S.L. ("Hauzed", "we") uses Artificial Intelligence (AI), automation and document processing on the website www.hauzed.com and the Hauzed mobile application (together, the "Platform").

This document forms part of our Privacy Policy and our Terms of Service. Where it adds detail, this Information about AI takes precedence on AI-specific matters. In the event of conflict, the Privacy Policy and Terms of Service prevail.

1. The principle in one sentence

Hauzed uses AI to support safer and more efficient rental workflows. Hauzed AI does not make final housing decisions.

This is architecture, not marketing language. Final decisions about accepting a tenant, signing a lease or releasing funds are always made by human beings (the landlord, the agent or the tenant) or by the contractual flow we define outside the AI system.

2. Hauzed's position under the EU AI Act

Regulation (EU) 2024/1689 (the "AI Act") classifies AI systems by risk. Annex III of the AI Act lists, among other things, "AI systems intended to be used to evaluate the creditworthiness of natural persons or establish their eligibility for essential private services". Housing is an essential service.

Our AI systems are designed as assistive tools that support the decision-maker. Because they do not autonomously determine eligibility, access or outcomes for tenants, they are not classified as high-risk AI systems under Annex III of the AI Act. We comply with the transparency obligations of Article 50 of the AI Act and the AI literacy obligation of Article 4. We do not use the prohibited practices in Article 5.

If regulatory interpretation evolves and any of our systems is reclassified, we will adapt our compliance and update this Information.

3. The AI-supported functions currently active at Hauzed

We currently use four AI-supported functions. We describe them functionally rather than by internal code name:

• Matching support - prioritises tenants or listings based on declared preferences, verification state, budget compatibility and other non-sensitive signals. The output is a suggestion, never a final decision.
• Message drafting support - suggests reply text for landlord or tenant conversations. Every message is reviewed and sent by a human.
• Lead aggregation - collects and organises externally forwarded leads into a central inbox view. It does not reply or decide.
• Social publishing support - schedules and publishes a listing only after the landlord has chosen to share it.

We do not publish internal prompts, scoring thresholds, routing logic or model configuration. Those details are part of our implementation and are kept internal for security and product integrity reasons.

We will keep this section updated when we add or remove AI-supported functions. When a materially new AI function is added, we will publish its role here at least 14 days before it goes live and notify registered Members.

4. AI-assisted features that are not independent agents

In addition to the agents above, we use AI in the following supporting features:

• Trust signals and badges - examples: "Identity Verified", "Income Evidence Submitted", "References Available", "Trusted Profile". They are derived from data the user provided plus structured document extraction from an enterprise provider. We do not show badges that imply negative or sensitive conclusions about a user (such as "low income" or "high risk"). We do not generate badges from protected characteristics.
• Matching relevance - when a tenant browses listings or a landlord browses applicants, the order is influenced by relevance signals (declared preferences, location, budget compatibility, verification status). The order is a suggestion; the user can sort or filter however they want.
• Fraud pattern detection - we use machine-learning models to detect suspicious behaviour (impossible action speed, duplicate listings, document inconsistencies). When a pattern is detected, Hauzed's human support team reviews the case before any action is taken on the account.
• Document understanding - described in Section 5.

5. Document processing pipeline

When you upload a document (payslip, employment contract, landlord reference, bank statement, study certificate, residence permit or similar):

1. Storage: the file is encrypted in transit (TLS 1.3) and stored encrypted at rest (AES-256) in our cloud storage.
2. Structured extraction: we send the file to a comparable enterprise document-processing service to extract structured fields (employer name, declared income range, contract dates, currency, etc.). The service converts unstructured documents into useful data fields.
3. Trust signals: we use the extracted data, together with other signals, to generate trust badges and matching relevance.
4. AI-assisted analysis (optional): where useful, we may use enterprise AI providers (such as Anthropic and OpenAI under commercial agreements) to generate additional signals or summaries. We never use these tools to make a final decision about a tenant.
5. Display: only the badges and signals are shown to the landlord. The original document is not shared with the landlord unless you choose to share it through chat.

What we do not do

• We do not put your documents into public general-purpose AI systems.
• We do not use your documents to train any AI model - neither ours nor a third party's. When we use enterprise AI providers, our agreements with them prohibit them from using customer data to train their general models.
• We do not use protected characteristics (race, ethnicity, religion, sexual orientation, disability, family status, gender identity) to generate signals, scores or matches. If the document processor incidentally extracts data of this nature, the affected fields are discarded at the extraction stage and are not stored.
• We do not generate signals such as "low income", "high risk", "high probability of default" or comparable negative labels.
• We do not do credit scoring.

6. Sharing documents through chat

When you, as a tenant, decide to share a document with a landlord through Hauzed chat, you do so voluntarily. Hauzed transmits the document; we do not require sharing as a condition of using the Platform.

From the moment the landlord receives the document, that landlord is an independent controller for that copy, and Hauzed does not control how the landlord stores or uses the document. We recommend sharing only the documents necessary for the rental application and hiding any sensitive fields you prefer not to disclose.

7. Your rights related to AI

In addition to the rights described in our Privacy Policy, when AI is involved you have the right to:

• Be informed that you are interacting with an AI agent rather than a human (Article 50.1 of the AI Act);
• Request an explanation of how a signal, score or recommendation about your profile was produced;
• Request human review of any signal or recommendation generated by AI, by writing to info@hauzed.com;
• Request rectification of any data extracted by the document processor that you believe is inaccurate (for example, if OCR misread a number);
• Object to AI-assisted processing based on legitimate interests (Article 21 GDPR);
• Not be subject to a decision based solely on automated processing that produces legal or similarly significant effects (Article 22 GDPR). Because we do not make final automated decisions about housing access, this article generally does not apply, but we respect the underlying rights as a minimum standard.

To exercise any of these rights, write to info@hauzed.com from the address associated with your account. We respond within 30 days.

8. Enterprise AI providers we work with

The external AI providers we currently work with are listed below. Each provider is bound by a Data Processing Agreement and Standard Contractual Clauses where applicable.

Provider	Service used	Data flow	Training restriction
Enterprise document-processing provider	Extraction of document fields	Document -> structured fields	Documents are not used to train the provider's general models
Anthropic (Claude API)	AI assistance under Enterprise / API agreement	Limited prompt context	Customer data is not used to train Anthropic's general models
OpenAI (API)	AI assistance under Enterprise / API agreement	Limited prompt context	Customer data is not used to train OpenAI's general models

We will update this list when we add or remove providers. The current list is also reflected in the sub-processor section of our Privacy Policy.

9. Security and supervision

Our AI systems are subject to:

• Logging: each signal and recommendation generated by an AI agent about a Member is logged and auditable, so we can respond to access and explanation requests.
• Bias review: we periodically review our matching models and signals to detect and correct possible bias against protected groups.
• Human oversight: a person on the Hauzed team is responsible for supervising the behaviour of each AI agent, with authority to disable or correct it at any time.
• Incident response: if we detect that an AI agent has produced harmful or biased output, we suspend the agent, investigate and inform affected users where appropriate.
• AI literacy: our team has documented training in AI literacy in line with Article 4 of the AI Act.

10. What you can ask us

If you believe that an AI signal, badge or recommendation about you is unfair, inaccurate or has unfairly affected your access to housing, write to info@hauzed.com. We will:

• Provide you with a plain-language explanation of how the signal was produced;
• Have a human review the signal and underlying data;
• Correct or delete the signal where appropriate;
• Update your profile and regenerate signals where applicable.

If you are not satisfied with our response, you can lodge a complaint with the Spanish Data Protection Agency (AEPD) at www.aepd.es or the Irish Data Protection Commission (DPC) at www.dataprotection.ie.

11. Changes to this Information

We may update this Information when we add, remove or change AI systems on the Platform. The "Last updated" date at the top reflects the most recent change. We will notify Members of material changes by in-app banner or email at least 14 days before they take effect.

12. Contact

For any question about how Hauzed uses AI:

GET HAUZED, S.L.
Calle Circumval·lació, 77, 3º 3ª, 08240 Manresa, Barcelona, Spain
Email: info@hauzed.com
Phone: +34 936 07 56 78

---

In the event of conflict between language versions, the Spanish version prevails.