Acceptable Use Policy

This Acceptable Use Policy (the “AUP”) governs your use of all products and services operated by Sylvanity B.V.under the Agentyk name — including the Agentyk chat product, our OpenAI-compatible APIs (such as Agentyk Scribe speech-to-text), the licensing and account platform at api.agentyk.xyz, the dashboard, any on-device software, and any distribution of these through a third-party marketplace (for example RapidAPI). It forms part of, and is incorporated by reference into, our Terms of Service. If you access Agentyk through an application built by someone else, both you and that developer are responsible for complying with this AUP.

Prohibited uses

You may not use any Agentyk product or API, and may not allow anyone using your account or integration to use it, for the following. This list describes categories of conduct; it is not exhaustive, and the absence of a specific example does not imply permission.

• Child sexual abuse material. Any content that sexualises, exploits, or endangers minors. Zero tolerance. Confirmed cases are reported to the competent authorities and relevant hotlines.
• Violence and weapons. Promoting, planning, or providing operational instructions for terrorism, mass violence, or physical harm to people or animals; the development, acquisition, or use of weapons, including chemical, biological, radiological, nuclear, or high-yield explosive weapons.
• Harassment and exploitation of people. Targeted harassment, threats, bullying, stalking, or defamation of identifiable individuals; non-consensual intimate imagery; doxxing or other unlawful exposure of private information.
• Fraud, deception, and abuse of systems. Fraud, scams, phishing, spam, plagiarism for academic or professional dishonesty, malware, ransomware, or any attempt to gain unauthorised access to, disrupt, or overload systems (including ours), or to circumvent our security, rate limits, licensing controls, or model entitlements.
• Privacy and unlawful surveillance. Collecting or processing personal data without a lawful basis; covert or unlawful recording, transcription, or monitoring of people without the consent or notice that the law requires; biometric identification, categorisation, or profiling of individuals without a lawful basis and appropriate safeguards.
• Practices prohibited by the EU AI Act. Regardless of anything else in our terms, you may not use Agentyk for any practice prohibited under Article 5 of Regulation (EU) 2024/1689, including social scoring, harmful manipulative or deceptive techniques, exploiting the vulnerabilities of specific groups, untargeted scraping of facial images to build recognition databases, emotion inference in workplaces or educational settings, and real-time remote biometric identification in public spaces.
• Misrepresentation and disinformation. Impersonating a person or organisation; presenting AI-generated output as human-authored where disclosure is legally required; or producing deceptive content at scale to manipulate elections, markets, or public opinion.
• Intellectual property and confidentiality. Infringing the intellectual property, trade-secret, or contractual rights of others, or submitting content you do not have the right to process.
• High-risk reliance without human oversight. Using outputs as the sole basis for decisions with significant legal, financial, safety, or health consequences for a person, or presenting outputs as professional medical, legal, or financial advice, without qualified human review.
• Reverse engineering and resale. Reverse-engineering, redistributing, or reselling our models or engines, or reselling raw API access, except as expressly permitted by your agreement with us.

Your responsibilities as a customer

If you build on our APIs, you are responsible for how your application and your own end users use the service. In particular, you agree that:

• You have a lawful basis and any necessary consents for the data you submit — especially audio, images, or other personal data — and you provide the notices the law requires to the people that data is about.
• You will not submit special-category data (for example health, biometric, or data revealing protected characteristics) unless you have an appropriate lawful basis to do so.
• You impose acceptable-use terms on your own end users that are at least as protective as this AUP, and you comply with the laws of your jurisdiction and those of your users.
• You remain responsible for reviewing outputs before relying on them, and for any decisions you or your users make based on them.

How we enforce this

We enforce this policy with the lightest mechanism that works for each product, and we will not send your content to a third-party moderation service to do it — that would betray the same privacy promise the rest of the product is built on.

• Chat.Each chat request first passes an automated safety check that runs on our own EU infrastructure, so your message never leaves Agentyk. We screen for a deliberately narrow set of hard-floor categories and refuse with an in-chat reply linking back to this page. We do not store the refused message — only the verdict and your account id.
• APIs (including Scribe). Audio and transcripts are processed in memory and never stored, so we do not inspect API content. Instead, API use is governed by this AUP and your agreement, and we monitor only operational metadata (request volume, error rates, abuse reports). Compliance is your responsibility as the integrator.
• Model-level safety. The language models are instruction-tuned with their own safety training and may decline individual requests at their own judgement, independent of the above.

Consequences of violations

Depending on severity, we may issue a warning, throttle or suspend access, or terminate the account or integration without refund. A single misjudged prompt is not grounds for termination; repeated or serious violations are. For content or conduct that poses an imminent threat to life, or that involves child sexual abuse material, we may suspend access immediately and report to the competent authorities, as required by law.

Reporting abuse and false positives

To report a violation of this policy, or if you believe a request was refused incorrectly, email hello@sylvanity.eu. For a refusal you can paraphrase what you were trying to do — there is no need to send the original prompt, since we will not have stored it. Automated classifiers make mistakes; we review reports and adjust scope where it is clearly miscalibrated.

Changes and governing law

We may update this AUP as our products, the law, or the threat landscape change. Material changes will be reflected in the “last updated” date above and, where appropriate, announced in-product. Continued use after a change means you accept the revised policy. This AUP is governed by the laws of the Netherlands and is subject to the same jurisdiction and liability terms as our Terms of Service.