Responsible AI Policy
How Wyvern uses artificial intelligence, and the limits that apply when personal information is involved.
Wyvern Innovations, also known as Wyvern Corp., designs software, automation, and AI systems for organizations in the Philippines. This policy says how the company approaches that work. It is a governance statement, not a certificate, and not a claim that a particular product has been audited by the National Privacy Commission or another authority.
The corporate website itself does not send visitor messages to an AI model. The Privacy Policy describes what the website does collect. Mark’s Health Grok Bot is a separate, unfinished assistant with its own notice at /marks-grok-bot.
1. Where AI is used
Wyvern offers AI as part of client projects: systems that explain information, sort work, or draft material for a person to review. A project uses AI only when that use is part of the agreed scope. The company will tell the client what the system is meant to do, what it should not be used for, and that its output can be wrong.
The website does not decide whether to hire, insure, credit, or treat a person. If a future system would make or recommend a decision with a legal or similarly significant effect on a person, that effect will be described before the system is used, and a person authorized by the controller will be able to review it. No such decision system is operating on this website.
2. Transparency and limits
AI output is generated text or a generated classification. It can omit context, copy an error in the source, or sound more certain than the evidence allows. Reference material, including a laboratory range or a legal citation, has to be checked against the original. An explanation produced with AI is not a professional opinion unless a qualified person adopts it.
Wyvern will not describe a conventional program as AI in order to make it sound more capable. NPC Advisory No. 2024-04 calls that practice AI washing when it harms a person whose data is processed. The company treats that advisory as applying whenever an AI system processes personal data, including during testing.
3. Human oversight
Staff read contact-form inquiries themselves. No model ranks or rejects those inquiries.
For a client system, consequential output is not a final decision by itself. Before a result is used to grant or deny a benefit, to diagnose, or to take a legal step, an authorized person reviews it. The client may take that review under the contract. A person affected by such a result may ask the controller to have a person look at it again. On this website, that request goes to hello@wyvern.com.ph. A written internal review procedure was not part of the website repository and still needs management approval before it is described in more detail.
4. Personal data and AI providers
Personal information is sent to an AI provider only when there is a lawful basis under the Data Privacy Act, the person or the client has been told, and the send is limited to what the task needs. This website does not make that send.
The company has not verified a standing rule that every AI provider excludes prompts from model training. Because that rule is not verified, the website does not submit inquiries, client files, or health information to a model. A client project will not send personal or confidential information to a provider until the contract states whether the provider may retain it or use it for training, and the client has agreed to that term.
Wyvern does not use contact-form messages to train its own model. The company does not sell prompts.
5. Fairness, errors, and misuse
Where an AI system processes personal data, Wyvern will look for systematic, human, and statistical bias that could treat people unfairly, and will limit the data used to what the stated purpose needs. Output that is wrong, misleading, or used to deceive is the responsibility of the organization that chooses to rely on it. Wyvern remains accountable for the processing it controls and for the processors it engages, which is the accountability rule in NPC Advisory No. 2024-04.
Clients and users may not use a Wyvern system to misrepresent AI output as a government, medical, or legal determination, or to make an automated decision about credit, insurance, employment, or housing unless a qualified person reviews it and the law allows it.
6. Client information
Documents a client provides for a project are used for that project. They are not added to a public training set. On the website, a client should not place confidential files in the contact form. The form stores the message and emails it to the company so a person can reply.
If Wyvern processes personal data for a client, the client decides the purpose and Wyvern follows documented instructions. If Wyvern decides the purpose, as it does for this website’s own inquiries, Wyvern is the controller. Which role applies to a project should be written in the contract. A standard processor agreement was not in the website records.
7. Philippine AI governance
Executive Order No. 131, signed on 12 November 2026, adopts the AI Governance Framework for the Philippines and creates the National Artificial Intelligence Council. The order directs government. It does not create a license or a penalty for private companies, and it leaves sector regulators with their existing mandates. News reports of the order describe principles of transparency, human-centric design, fairness, accountability, security, and safety. Wyvern uses those principles as guidance. The company does not claim that the order has been independently certified against Wyvern’s products, and the official gazette text was not attached to this website.
NPC Advisory No. 2026-01 governs scraping of publicly available personal data. This website does not scrape. A project that would scrape personal data has to meet that advisory before it starts, including a declared purpose, a lawful basis, notice, and a privacy impact assessment. Public availability is not consent.
8. Changes and contact
This policy will be updated before the website begins sending personal information to an AI provider. Questions go to Wyvern Innovations, 12/F Times Plaza Building, United Nations Avenue, Ermita, Manila, at hello@wyvern.com.ph.