Artificial Intelligence System Disclaimer

Version: 2.1 — Updated October 9, 2026. Scope: The entire multi-agent AI assistant system on the LandAIA platform, including all operating modes.

1. Nature and technical limitations of AI

The User acknowledges and agrees that the Artificial Intelligence ("AI") features on LandAIA are assistive tools based on Large Language Models, statistical and probabilistic algorithms. They are not human and do not have full awareness of specific legal, factual, or transactional contexts.

Due to technical limitations, AI may:

• Produce information that is inaccurate, incomplete, outdated, or unsuitable for a specific case (AI Hallucination)
• Generate predictive, inferential, or reference-only results that are not firm commitments or conclusions
• Misinterpret the user's context, intent, or requests, leading to incorrect actions

LandAIA does not guarantee and is not liable for the absolute accuracy of AI-generated content or actions.

2. Multi-agent specialized system

2.1 System architecture

LandAIA operates on a multi-agent specialized AI system, in which multiple AI assistants ("Agents") collaborate to process user requests. Each Agent has its own specialization, including but not limited to:

• Orchestration & Intake: receives requests, analyzes context, dispatches to the right specialist
• Customer Management (CRM): creates, updates, deletes customer information; sends messages; recommends properties
• Property Inventory: searches, filters, and matches properties in the system
• Maps & GIS: analyzes location, planning, surrounding areas
• Legal: looks up legal information on real estate (reference only)
• Valuation & Market: analyzes prices and market trends
• Field Survey: creates, updates, deletes surveys; edits property data
• Content & Communication: generates listings, property descriptions
• Image Processing: analyzes and labels property images
• Storage & Memory: records and summarizes conversation context
• Multi-department Orchestration: breaks down and distributes complex requests

2.2 Specific characteristics

The User acknowledges and understands that:

• The list, names, functions, and number of Agents may change with each system update without prior notice
• Agents collaborate automatically with each other — a single user request may be processed sequentially or in parallel by multiple Agents
• Each Agent has access limited to its assigned tools, but may still cause unintended effects within that scope
• Any action taken by any Agent in the system is deemed an action of the "LandAIA AI system" — the User may not rely on "Agent was wrong" or "Agent was not right" as a reason to avoid responsibility

3. Operating modes and AI intervention levels

The system provides three (03) operating modes with different levels of AI autonomy. Users select the appropriate mode in application settings:

3.1 Special terms for Autonomous mode

When a User voluntarily enables Autonomous mode:

• Every action performed by AI (including creating, modifying, permanently deleting data, sending messages/content externally) is deemed the direct will and act of the User
• The User has voluntarily waived the right to pre-review and the right of intermediate confirmation
• LandAIA and the operating entity (MINH THINH TECH GROUP JSC) are, to the maximum extent permitted by law, released from liability for any losses, errors, or damages arising in this mode

3.2 In-flight actions (Atomic Operations)

When the User switches from Autonomous mode to another mode:

• Actions already initiated and being executed by AI at the moment of switching will still complete
• Only new actions after the switch are subject to the new mode
• The User remains responsible for such in-flight actions

3.3 Separate authorization agreement

Autonomous mode requires the User to sign a separate Authorization Agreement (Autonomous Consent) before activation. This Agreement includes detailed provisions regarding authorization, liability waiver, third-party responsibility, and emergency kill-switch mechanism. The content of the separate Authorization Agreement is a supplementary and inseparable part of this AI Disclaimer.

4. AI is not a substitute for professional advice or competent authorities

Content, suggestions, or analyses generated by AI (including results from any Agent in the system) do not substitute for:

• Legal, financial, investment, or valuation advice from certified professionals
• Planning and legal information on real estate from competent State authorities
• Inspection of records, documents, current status, and dispute status of real estate

All information related to prices, planning, legal matters, and investment potential is for reference only at the time of query. Even if an Agent specializing in legal matters or valuation provides analysis, the result is still for reference only and has no legal validity.

5. Input data and security responsibility

5.1 Sensitive data

Users are advised not to provide the following sensitive data to the AI system:

• National ID/citizen ID numbers
• Bank accounts
• Passwords, OTP codes
• Trade secrets or personal data of third parties

LandAIA disclaims all liability for risks arising if Users voluntarily provide such information. Note: In the multi-agent system, information You provide may be shared between Agents to process Your request.

5.2 Lawfulness of data

The User is solely responsible for:

• The lawfulness of data and content submitted to AI
• Ownership or lawful rights to use such data
• The accuracy of customer and property data that AI uses to perform automated actions

6. Intellectual property in AI-generated content

• LandAIA permits Users to use AI-generated content for personal purposes or internal business support
• LandAIA does not warrant that Users hold exclusive intellectual property rights or authorship in such content under applicable IP laws
• Use of AI content must comply with Vietnamese law and must not infringe third-party rights

7. Maximum disclaimer

To the maximum extent permitted by law, LandAIA, the operating entity (MINH THINH TECH GROUP JSC), shareholders, partners, and operations team are not liable for:

• Financial losses, missed investment opportunities, or disputes arising from Users' use of or reliance on AI content
• Information inaccuracies due to changes in legal documents, planning, or State policies
• System errors, interruptions, or limitations arising from infrastructure providers, AI models, or third-party services
• Any erroneous actions, data created/modified/deleted in error due to AI misinterpreting intent or hallucination, especially in Autonomous and Semi-autonomous modes
• Any messages or content automatically sent to customers or third parties that do not reflect the User's actual intent

8. User responsibility towards third parties

The User is fully liable before the law and to third parties (customers, partners, regulators) for:

• Any content, commitments, or transactions initiated or performed by AI on behalf of the User
• Any CRM messages, property recommendations, or information sent to customers through the AI system

LandAIA bears no joint responsibility in any dispute arising from AI's automated actions.

9. Mandatory user obligations

By using AI features, the User commits to:

• Independently verify information from authoritative and independent sources
• Not use AI to create fraudulent, deceptive, or unlawful content
• Take sole responsibility for all real estate transaction decisions
• Select the operating mode that matches the level of oversight they desire
• Actively disable Autonomous mode upon detecting abnormal signals from the system

10. Compliance with AI laws

Consistent with the Law on Artificial Intelligence No. 134/2025/QH15, the Law on Digital Technology Industry No. 71/2025/QH15 and their implementing documents:

10.1 Transparency when interacting with AI

• Users are always informed that they are interacting with an AI system; assistants are named and presented as digital assistants and never impersonate real people
• Human control is ensured through the three-mode mechanism, in which Users decide the level of AI autonomy
• The system fully records AI action history to support inspection and accountability

10.2 Labelling and verification codes for AI-generated content

Content drafted by AI assistants for delivery to third parties through the Platform — customer-care messages sent via the Customer Portal and listing descriptions on enterprise websites (not listing titles) — carries identification marks in two layers:

• A visible label with a verification code in the form #AI-xxxxx, stating that the content was drafted by a LandAIA digital assistant;
• An invisible signature embedded in the text using non-printing characters, allowing traceability even if the visible label is stripped during copying.

For AI-drafted content that users post themselves on external platforms (listing sites, social networks…), LandAIA does not insert a label or invisible characters into the content (receiving platforms may reject unusual characters): the verification code is recorded in LandAIA's records and shown to the user so they know which parts were drafted by AI; users state clearly that the content was produced with AI assistance when posting. The labelling scope continues to expand to other content-generation features under LandAIA's compliance roadmap.

For images generated or edited by AI

Images generated or edited by LandAIA's AI carry three layers of identification:

• Visible mark: that image's lookup code — in the form AI-XXxxxx — printed at the bottom-right of the image; the mark is semi-transparent and does not obscure image content. Images published before the visible-mark update carry the form AI · AI-XXxxxx; the system does not re-stamp already-published images, so both forms remain in circulation. When looking a code up, enter it in full, including the AI- prefix;
• Invisible watermark embedded in the image data;
• C2PA metadata recording content provenance under the international content-provenance standard.

10.3 Public verification page

Technical limitations to note:

• For text: the invisible signature may be lost if content is retyped manually or copied through applications that strip special characters;
• For images: this comparison relies on a file integrity hash, so it only matches while the file is intact. Images compressed or altered by messaging apps and social platforms will not match;
• In both cases, use the verification code printed with the content or shown on the image — the most stable basis for lookup.

10.4 User obligations regarding AI labels

• Do not remove, conceal, or falsify labels, verification codes, or invisible signatures applied by the Platform to AI content;
• When publishing or distributing AI-generated content outside the Platform, Users are responsible for preserving the label or fulfilling the labelling obligation under applicable law;
• Do not use AI content to impersonate real people, create false information, or mislead customers and third parties.

11. General provisions

LandAIA reserves the right to adjust, limit, or suspend AI features (including adding, removing, renaming, or altering functions of Agents) within the scope permitted by law, for system maintenance, service upgrades, or compliance with new State regulations.

When this Disclaimer is updated, the new version will be published at the same URL. Users currently in Autonomous mode will be required to reconfirm the authorization agreement under the new version before continuing to use that mode.