OULUTION LEGAL
OULUTION AI TRANSPARENCY NOTICE
ON THIS PAGE
- Why we publish this Notice
- 1. You are interacting with an AI system
- 2. What Oulution does and does not do
- 3. How Oulution generates responses
- 4. Known limitations
- 5. Your data and AI model training
- 6. Human oversight
- 7. Automated decision-making
- 8. Risk classification under the EU AI Act
- 9. Provider and deployer roles
- 10. Copyright, training data, and content used to build the models
- 11. Safety, content policy, and abuse
- 12. Interaction with law enforcement and regulators
- 13. Contact
- 14. Changes to this Notice
Why we publish this Notice
This Notice explains, in plain language, how Oulution’s artificial intelligence works, what it can and cannot do, and how you can control your data. It is issued by Oulution Holdings Limited and is designed to satisfy transparency obligations under Article 50 of the EU Artificial Intelligence Act (Regulation (EU) 2024/1689), equivalent transparency expectations under the UK, US state, Hong Kong, People’s Republic of China, and other emerging AI-governance frameworks, and to give you the information you need to use the Services responsibly.
Read this Notice together with the Oulution Terms of Service (Consumer or Enterprise, as applicable), the Privacy Policy, the Cookie Policy, and the Acceptable Use Policy. Defined terms used but not defined here have the meaning given in the applicable Terms of Service.
1. You are interacting with an AI system
1.1 When you use Oulution, you are interacting with an artificial intelligence system. Oulution generates responses using large language models combined with retrieval from Oulution’s curated legal database. Responses are produced by pattern-based prediction, not by legal reasoning of the kind performed by a qualified lawyer engaged in a professional retainer.
1.2 You are not interacting with a human lawyer. No lawyer-client relationship, solicitor-client relationship, or attorney-client relationship is formed by using the Services. Oulution is not a law firm and does not hold a legal practising certificate in Hong Kong or in any other jurisdiction.
1.3 This Notice is a standing disclosure. In addition, at appropriate points inside the product interface, Oulution surfaces a short in-line disclosure that you are interacting with an AI system. That in-product disclosure and this Notice together are intended to discharge Oulution’s transparency obligations to individual users under applicable AI-governance law.
2. What Oulution does and does not do
2.1 Oulution provides:
(a) legal information and analysis on statutes, regulations, treaties, and legal concepts;
(b) comparative summaries of legal frameworks across jurisdictions;
(c) issue-spotting and risk flags on transactions and documents you submit;
(d) structured checklists, diligence templates, and analytical frameworks;
(e) document review, drafting assistance, and workflow support;
(f) introductions, on request, to independent qualified lawyers.
2.2 Oulution does not:
(a) give legal advice tailored to your specific situation in the manner of a lawyer engaged on a retainer;
(b) certify that a course of action is legally compliant;
(c) issue formal legal opinions within the meaning of any professional-conduct rule;
(d) represent you before any court, tribunal, regulator, or counterparty;
(e) take professional responsibility for outcomes arising from the use of Outputs.
2.3 Where a matter requires legal advice, you should consult a qualified lawyer admitted in the relevant jurisdiction. Where Oulution introduces you to a lawyer through its lawyer-match feature, any resulting engagement is a direct contract between you and that lawyer to which Oulution is not a party.
3. How Oulution generates responses
3.1 Oulution combines the following components:
(a) large language models operated by Oulution and/or accessed via contracted third-party model providers under no-training contractual terms;
(b) retrieval from Oulution’s curated legal database, comprising statutes, regulations, case law, regulator guidance, and secondary sources;
(c) structured prompts and system instructions designed to maintain quality, consistency, and jurisdictional discipline;
(d) safety, content, and jurisdictional filters.
3.2 Because the underlying models are probabilistic:
(a) responses may vary between sessions even for identical questions;
(b) responses may be incorrect, incomplete, or out of date;
(c) statements may be presented with a level of confidence that is not warranted by the underlying accuracy of the model;
(d) the model may generate references, citations, or quotations that do not exist (“hallucination”).
3.3 You must not rely on any Output without independent verification against primary sources and, where the matter is material, without review by a qualified lawyer.
4. Known limitations
The following limitations are inherent to large-language-model systems as at the effective date of this Notice:
(a) Hallucination. The models may fabricate citations, case names, statutory references, quotations, or facts. Every citation must be independently verified.
(b) Training cut-off dates. The models are trained on data up to a specific date and may not reflect recent legal developments unless the retrieval sources are current at the time of the query.
(c) Jurisdictional accuracy. The models can conflate legal systems — most commonly confusing Hong Kong law with People’s Republic of China law, English law with United States law, or EU directives with the national implementations of Member States. When jurisdiction matters, ask explicitly and verify.
(d) Numerical calculations and dates. Numerical accuracy is not a strength of these models; independently verify any calculation, date, or quantitative statement.
(e) Confidential and privileged material. Outputs derived from your Inputs are only as reliable as those Inputs. Do not treat Outputs as legal advice or as certified analysis.
(f) Cross-language variance. Where you interact with the Services in a language other than English, quality of legal analysis in that language may vary from equivalent English-language responses.
5. Your data and AI model training
5.1 Consumers (individuals using the Services for personal purposes). By default, Oulution uses your Inputs and Outputs to train, evaluate, and improve its AI models under the legitimate-interests legal basis (GDPR Art. 6(1)(f)) and equivalent bases under other applicable data-protection laws. You may opt out at any time through your account settings. On opt-out:
(a) your future Inputs and Outputs will be excluded from training datasets;
(b) where technically feasible, your prior Inputs and Outputs will be excluded from datasets used to train new models after your opt-out;
(c) Oulution will not delete or retrain models that have already been trained on your prior Inputs before your opt-out — full retraining of a large model on request is not technically possible, and this limitation is recognised by leading data-protection regulators.
5.2 Enterprise customers. Oulution does not use enterprise Customer Data to train, tune, evaluate, or improve its AI models unless the enterprise customer expressly opts in through its Order Form or Master Subscription Agreement.
5.3 Excluded content — never used for training regardless of tier. The following categories are excluded from training datasets by design:
(a) content marked or reasonably identifiable as subject to legal professional privilege or the equivalent under any jurisdiction;
(b) special-category personal data (GDPR Art. 9), sensitive personal information (CCPA § 1798.140(ae)), or sensitive personal information (PIPL Art. 28);
(c) personal data of children;
(d) content flagged by the user as confidential and excluded from training.
5.4 Further detail on data processing, legal bases, retention, cross-border transfers, and your rights is set out in the Privacy Policy, which is incorporated into this Notice by reference.
6. Human oversight
6.1 Every substantive Output should be reviewed by a competent human before use. For anything material — a filing, a signed document, advice given to a third party, a regulatory submission, a transaction — that human should be a qualified lawyer admitted in the relevant jurisdiction.
6.2 The Services are designed to support qualified human decision-making, not to replace it. Any use of the Services in a manner that treats Outputs as final legal work-product without human review is inconsistent with this Notice and, depending on the jurisdiction, may constitute the unauthorised practice of law.
7. Automated decision-making
7.1 Oulution does not make decisions that produce legal or similarly significant effects on any individual within the meaning of GDPR Art. 22, UK GDPR Art. 22, or equivalent provisions of PDPO, PIPL, LGPD, or other applicable data-protection laws. Outputs are advisory in character and require your review before any action is taken.
7.2 If Oulution introduces features in the future that involve automated decision-making producing legal or similarly significant effects on individuals, those features will be preceded by a specific disclosure and, where required, obtaining consent or providing opt-out rights consistent with applicable law.
8. Risk classification under the EU AI Act
8.1 Oulution’s assessment, based on the EU Artificial Intelligence Act (Regulation (EU) 2024/1689) as it enters into phased application, is that the Services are a limited-risk AI system subject to the transparency obligations of Article 50, which this Notice is designed to discharge.
8.2 In Oulution’s assessment, the Services are not a “high-risk” AI system under Annex III of the AI Act. That assessment rests on the following features of the Services:
(a) Outputs are legal information and analysis for human review, not automated legal determinations;
(b) Oulution does not deploy the Services to administer justice or democratic processes;
(c) Oulution does not use the Services to make eligibility, entitlement, or scoring decisions concerning natural persons.
8.3 Oulution keeps this classification under continuous review. If the classification changes as a result of feature development, regulatory guidance, or judicial interpretation, this Notice will be updated and, where required, the additional obligations attaching to a higher-risk classification (including conformity assessment, registration in the EU database, and post-market monitoring) will be implemented.
9. Provider and deployer roles
9.1 In the terminology of the EU AI Act, Oulution is the provider of the Services. Where a customer (typically an enterprise) integrates the Services into its own workflow and puts them into service under its own responsibility for its own users, that customer may be a deployer and will have its own obligations under Article 26 of the AI Act.
9.2 Oulution’s Enterprise Terms of Service and Master Subscription Agreement contain the technical documentation, instructions for use, and cooperation commitments that a deployer requires to discharge its own obligations. Enterprise customers who require the technical documentation package for their AI-governance file may request it at info@oulution.ai.
10. Copyright, training data, and content used to build the models
10.1 Oulution takes intellectual-property compliance in AI training seriously. Where Oulution trains or fine-tunes its own models, it uses:
(a) data that Oulution has the right to use (including public-domain material, appropriately licensed material, material published by governments and public authorities as legal source material, and data lawfully collected from users under the legal bases described in the Privacy Policy);
(b) data provided by contracted third-party model providers whose foundation models Oulution accesses under commercial terms addressing training data provenance.
10.2 Oulution respects text-and-data-mining opt-outs expressed under Article 4 of Directive (EU) 2019/790 (the DSM Directive) and equivalent rights under other applicable copyright frameworks, where such opt-outs are expressed in a machine-readable form on the source material.
10.3 If you believe that content in which you hold rights has been used to train Oulution’s models without authority, please contact info@oulution.ai with sufficient detail for Oulution to investigate.
11. Safety, content policy, and abuse
11.1 Oulution operates content filters and safety systems designed to prevent generation of unlawful, harmful, or manifestly abusive Outputs. No such system is perfect. If you receive an Output that you believe is inaccurate, harmful, unlawful, or that has caused you concern, please contact info@oulution.ai.
11.2 Oulution investigates all substantiated reports and uses the findings to improve safety, quality, and the underlying models.
11.3 Attempts to circumvent safety systems, jailbreak the Services, extract system prompts or model weights, or use the Services to generate content that breaches the Acceptable Use Policy are prohibited and may result in suspension or termination.
12. Interaction with law enforcement and regulators
12.1 Oulution complies with lawful requests from competent authorities. The framework for handling such requests is set out in the Privacy Policy (for personal data) and the Data Processing Addendum (for enterprise Customer Data).
12.2 Oulution will publish, at least annually, a summary of the number and nature of government access requests received, to the extent lawful.
13. Contact
Questions about this Notice may be sent to info@oulution.ai. Written correspondence may be sent to the registered office identified at the top of this Notice.
14. Changes to this Notice
Oulution will update this Notice as its AI systems and the applicable legal framework evolve. Material updates will be notified through the Services and will trigger a fresh version-and-date stamp at the top of this Notice.