CyberLink Responsible AI Statement

Ethically and Responsibly Harnessing the Power of AI

Last updated: September 2026

CyberLink may update this statement from time to time as our products, technologies, industry standards, and applicable laws evolve.

Our Commitment

CyberLink has worked with artificial intelligence for many years across multimedia creation and facial recognition. As AI becomes more capable and more widely used, our responsibility is to design, integrate, build, and deliver AI technologies in a way that is useful, trustworthy, and respectful of the people who use our products and the people who may be affected by them.

We aim to develop AI features that keep people in control, provide appropriate clarity about how AI is used, protect privacy and security, respect the rights of others, and evolve with applicable laws, industry standards, and responsible AI practices.

How CyberLink Uses AI

This statement applies to CyberLink products and services that include AI features. These may include digital media and creative tools, online creation services, marketing and design tools, and facial recognition and computer vision technologies.

The specific AI features available may vary by product, platform, region, subscription plan, and release. Not all products or features include generative AI, and not all AI features operate in the same way.

Our Responsible AI Approach

CyberLink's approach to responsible AI is guided by the following principles:

Human control and oversight. We design AI to assist users, customers, and organizations, not to remove human judgment from important decisions.

Clarity about AI use. We aim to help users and customers understand when AI is used and, where applicable, when third-party AI models or services are involved.

Privacy and security. We work to protect personal data and product security through appropriate technical and organizational measures, taking into account the nature of each product and deployment model.

Respect for rights. We design and operate our AI features with respect for privacy, intellectual property, portrait, personality, voice, and other individual rights in mind, in line with CyberLink's broader commitment to respecting internationally recognized human rights.

Safeguards against misuse. We apply usage policies and, where available, technical safeguards intended to reduce misuse of our AI features and the creation or use of unlawful, deceptive, infringing, or harmful content.

Responsible Generative AI and Creative Tools

Generative AI and AI-assisted creative tools can help users create, edit, enhance, and transform images, video, audio, text, and other content more efficiently. CyberLink aims to provide these capabilities in ways that support creativity while addressing transparency, rights, safety, and misuse risks. Generative AI models can also reflect biases present in the data they were trained on, and may produce results that are stereotyped or that do not represent people fairly. We take this into account when selecting and integrating the models we offer.

Creative control. Generated and AI-edited results appear in the editing environment, where you can review, adjust, replace, or discard them before export or publication, and CyberLink does not publish content on your behalf. The specific controls available vary by product and workflow. Generative AI can produce results that are inaccurate, unexpected, or unsuitable for a particular purpose, and outputs should not be relied on as statements of fact; we encourage users to review generated content before using or publishing it.

Rights, consent, and misuse safeguards. CyberLink's terms prohibit the use of our tools to create unlawful, infringing, deceptive, or harmful content, or to infringe the intellectual property rights of others. Users are responsible for ensuring they hold the rights needed for any material they provide to our AI features. We also provide reporting channels for suspected misuse, as described below.

Synthetic media and deepfakes. Our tools can produce realistic synthetic images, video, and audio, including features that involve a person's face or voice. Our terms require that you have the rights and consents needed for any person who appears in, or whose voice is used in, the material you provide, and prohibit using our tools to create deepfakes that impersonate real people for deceptive purposes, non-consensual intimate imagery, or content that infringes personality, portrait, voice, or privacy rights. We have zero tolerance for child sexual abuse material and for any use of our tools involving the sexualization of minors. We act on reports of such misuse and may suspend or terminate accounts.

Data, training, and third-party AI models. Product-specific terms and privacy disclosures describe how user content is processed and retained in connection with AI features. Some AI features are provided using third-party AI models or services; where that is the case, content submitted to those features is processed by that provider under its own terms. We assess third-party AI providers before integrating their models into our products. We do not use content you create or upload when using our AI features to train CyberLink's own generative AI models.

AI content identification. CyberLink believes people should be able to understand when digital content has been generated or materially modified by AI. CyberLink is committed to implementing and maintaining appropriate machine-readable content provenance and transparency measures for AI-generated and AI-modified content, which may include metadata, content credentials, digital watermarking, or other provenance technologies, where available and appropriate.

These measures may help record information such as whether content was generated or modified using AI, and may include information about the tool, service, or other provenance details involved. Availability may vary by product, feature, region, and release. Provenance metadata and similar technical measures are not a complete solution; they may be affected by downstream editing, re-encoding, file conversion, platform processing, or other actions outside CyberLink's control.

Applicable laws and platform rules may impose different marking or disclosure obligations depending on how AI-generated or AI-modified content is created, provided, used, or published. Those obligations may apply to CyberLink, users, or both, depending on the relevant product and use case. Users remain responsible for any disclosures required in connection with their own use or publication. Separately, and beyond what the law requires, we encourage users to disclose AI-generated or AI-modified content wherever the context calls for it.

Responsible Facial Recognition and Computer Vision

FaceMe® and its related computer vision technologies are CyberLink's facial recognition and AI inference solutions for business customers and other organizations, deployed and run within each customer's own environment. We believe these technologies can meaningfully advance real-world identity and security outcomes when designed and deployed responsibly, and we are committed to upholding rigorous standards for privacy, security, fairness, and respect for human rights throughout FaceMe®'s development.

FaceMe® is intended to support identity verification, access control, safety and security, and fraud reduction use cases.

Facial recognition accuracy can vary depending on factors such as image quality, camera placement, lighting, configuration, and deployment conditions. CyberLink evaluates FaceMe® through internal testing and independent third-party evaluation, including benchmarking against NIST Face Recognition Technology Evaluation (FRTE) results and iBeta Presentation Attack Detection (PAD) Level 2 certification for liveness and anti-spoofing performance, and evaluates fairness and demographic performance as part of model development, working to minimize performance differences across demographic groups. No facial recognition or anti-spoofing technology is error-free, and business customers should consider these limitations when designing and operating their deployments.

FaceMe® is designed to support deployment in environments controlled by the customer — whether on the customer's own servers, on-premises infrastructure, or the customer's own cloud environment. In these customer-controlled deployments, FaceMe® operates entirely within that environment, and CyberLink does not receive, access, or store the personal data or facial templates processed there. CyberLink does not operate a cross-customer global database of faces.

CyberLink provides product documentation, along with privacy and security guidance, to support business customers in deploying and operating FaceMe® responsibly. While customers control how FaceMe® is deployed within their own environments, they remain responsible for determining the lawful basis for their use, providing required notices, obtaining consent where required, honoring applicable individual rights, and complying with applicable laws. Regardless of how FaceMe® is deployed, CyberLink does not support uses of FaceMe® that are unlawful, discriminatory, deceptive, abusive, or inconsistent with internationally recognized human rights. Where FaceMe® is used to support decisions that have a significant effect on individuals, we also expect meaningful human review before such a decision is made.

For additional discussion of responsible facial recognition practices and related privacy considerations, please refer to our FaceMe® white paper: Responsibly Using Facial Recognition in the U.S. This white paper is focused on the United States; legal requirements differ in other jurisdictions.

Privacy, Reporting, and Responsible Use

CyberLink handles personal data that it directly collects or processes in accordance with its Privacy Policy and applicable laws. Where CyberLink directly collects or processes personal data, users may request deletion of their personal data as described in the Privacy Policy.

Our services are intended for users who meet the minimum age requirements set out in our Terms of Service.

If you believe our AI tools or products have been misused, or if you have questions about this statement, please contact us at privacy@cyberlink.com. CyberLink reviews reports of misuse and may suspend or terminate accounts for violations of our terms.

Governance and Accountability

Responsibility for AI at CyberLink is shared across our product, engineering, legal, and privacy functions. We consider privacy, security, rights, and misuse risks as part of developing and releasing AI features, and we update our internal policies, product terms, and disclosures as our capabilities change.

Our Ongoing Commitment

AI technology, industry standards, and regulation are evolving quickly. CyberLink will continue to monitor relevant requirements and responsible AI practices — including developments related to AI transparency, biometric data, privacy, security, content provenance, and platform rules — and to review and update our AI practices, product disclosures, safeguards, and this statement accordingly.

For more information, please refer to our Global Human Rights Principles, Privacy Policy, and Terms of Service.