Acceptable Use Policy

01General rules

You may not use the Service, or help anyone else use it:

• for anything illegal, or to promote, plan or carry out illegal activity, fraud, scams, phishing, harassment, stalking, threats, hate or violence against people;
• to generate or distribute child sexual abuse material or any sexual content involving minors — such content is reported to the authorities;
• to create malware, ransomware or exploits for systems you are not authorised to test, to gain unauthorised access to systems, accounts or data, or to run denial-of-service attacks;
• to develop or use biological, chemical, nuclear or radiological weapons, or other weapons intended to cause mass harm;
• to infringe intellectual property, privacy or other rights, or to process personal data without a lawful basis;
• to send spam, run deceptive bulk messaging or impersonate a person or organisation without authorisation;
• to automate a device, app or account you do not own or are not authorised to control, or to bypass the security, anti-automation or anti-abuse protections of any third-party service;
• to generate disinformation intended to deceive, including in elections, or deepfakes of real people without their consent;
• to clone, imitate or synthesise the voice, face or likeness of a real person other than yourself without that person’s prior written consent and our express permission, including public figures, or the voice or likeness of anyone under 18;
• to use synthetic voices or avatars to deceive people about who is speaking, including to obtain money, payment details, passwords or one-time codes, or to bypass voice- or face-based authentication (for example a bank’s voice ID);
• to create political or electoral content with a synthetic voice or likeness presented as authentic, or content that impersonates a candidate, official or public authority;
• to remove or hide the AI-generated label or the machine-readable provenance marks on audio or video made with Kodwen.

02Prohibited AI practices (EU AI Act, Article 5)

You may not use the Service to place on the market, put into service or use an AI system for any practice prohibited by Article 5 of Regulation (EU) 2024/1689, including:

• subliminal, manipulative or deceptive techniques that materially distort behaviour and cause significant harm;
• exploiting vulnerabilities of people due to their age, disability or social or economic situation;
• social scoring of people leading to detrimental or disproportionate treatment;
• assessing or predicting the risk that a person will commit a crime based solely on profiling or personality traits;
• creating or expanding facial-recognition databases through untargeted scraping of images from the internet or CCTV;
• inferring emotions of people in the workplace or in education institutions (except for medical or safety reasons);
• biometric categorisation to deduce race, political opinions, trade-union membership, religious or philosophical beliefs, sex life or sexual orientation;
• real-time remote biometric identification in publicly accessible spaces for law enforcement, outside the narrow exceptions of the AI Act.

Using Kodwen inside a high-risk AI system (Annex III of the AI Act) requires a written agreement with us first.

03If you build on Kodwen (deployer duties)

• Tell people they are interacting with an AI system unless it is obvious from the context (AI Act Article 50(1) as applied to your product).
• If your product generates or manipulates images, audio or video that resemble real people, objects, places or events (deepfakes), disclose that the content is artificially generated or manipulated (Article 50(4)).
• If you publish AI-generated text to inform the public on matters of public interest, disclose that it was AI-generated unless it was reviewed by a human who holds editorial responsibility (Article 50(4)).
• Keep any machine-readable AI-provenance markings that outputs carry; do not remove watermarks such as Google SynthID or C2PA manifests on generated images, audio and video.
• Give your own users terms at least as protective as this policy and the provider policies below.

04Model and service integrity

• Do not use outputs of the Service to develop, train or improve AI models that compete with Kodwen or with the model providers behind it (no distillation), and do not systematically harvest outputs for that purpose.
• Do not attempt to extract model weights, system prompts or routing logic, or reverse-engineer the gateway.
• Do not resell, sublicense, share or publish API keys or access to `codai` without a written agreement. One account per person; team seats as stated in your plan.
• Do not exceed rate limits, circumvent spend caps or interfere with the Service or other customers.

05Model provider policies

Requests to `codai` are fulfilled by third-party models. Their usage policies apply to your use of the Service as if you had accepted them directly, and a request that would violate them may be refused:

• Anthropic Usage Policy — https://www.anthropic.com/legal/aup
• Google Generative AI Prohibited Use Policy — https://policies.google.com/terms/generative-ai/use-policy
• Microsoft Enterprise AI Services Code of Conduct — https://learn.microsoft.com/en-us/legal/ai-code-of-conduct
• xAI Acceptable Use Policy — https://x.ai/legal/acceptable-use-policy

06Sandboxes, code execution and preview links

Kodwen sandboxes run code you or your agents send, in isolated containers with network access only to the domains you allow. In addition to the rules above, you may not use sandboxes:

• to mine cryptocurrency, run proof-of-work or keep resources busy for anything other than the task you submitted;
• to scan, probe, brute-force or attack any network, host or service, or to try to reach Kodwen internal systems, cloud metadata services or other customers’ sandboxes, or to escape or weaken the isolation;
• as a proxy, VPN, relay or anonymiser, to send email or bulk traffic, or to download or redistribute content you have no right to;
• to host or serve through a preview link phishing pages, malware, illegal content or anything else that breaks this policy; a preview link is short-lived and meant for you and the people you choose to show your work to.

Content reachable through a preview link is content you host on codai: you are responsible for it, and we act on notices about it as described at codai.ro/dsa. We may stop or delete a sandbox, its snapshots and its files, and revoke its preview links, when we reasonably believe it breaks this policy or puts the Service or others at risk.

07Enforcement and reporting

We may refuse requests, suspend keys or accounts, or terminate access if we reasonably believe this policy is being broken, and we may report illegal content or activity to the competent authorities. Where possible we tell you why and give you a chance to respond, unless that would be unlawful or create a risk of harm.

Report abuse to hello@codai.ro or security@codai.ro; for illegal content hosted on Kodwen (for example a shared transcript or Studio creative), see codai.ro/dsa.