AI text rewriting acceptable use
You may inspect and clean text you created, client or team content you are authorized to edit, public-domain material, and material licensed for the intended transformation. Legitimate purposes include repairing encoding, adapting approved copy, privacy review, creating a fresh draft, and integrating the Scale API into an authorized publishing workflow.
You must review output and honor attribution, confidentiality, disclosure, accessibility, and recordkeeping duties that apply to the work.
Prohibited deception
Do not use the service to misrepresent academic or professional work, evade a required AI disclosure, plagiarize, impersonate another person, conceal stolen material, commit fraud, fabricate evidence, or induce someone to rely on false provenance. Rewriting does not turn unauthorized source material into authorized work.
Do not describe NoLLMWM as an official Anthropic, Google, OpenAI, or other vendor verifier. Do not claim guaranteed watermark removal when vendor verification is unavailable.
Prohibited harmful content and system abuse
Do not process content to facilitate malware, credential theft, targeted harassment, exploitation, illegal discrimination, or other unlawful harm. Do not probe other accounts, bypass quotas, share API keys publicly, automate abusive signups, overload endpoints, replay payment events, or attempt to extract server secrets.
Security research requires good-faith limits: use your own account and data, avoid service disruption, and report findings privately through the security contact.
Enforcement and reporting
We may rate-limit, refuse, suspend, or terminate activity that presents abuse, legal, payment, or security risk. We may preserve minimum records needed to investigate an incident and comply with law. Enforcement does not create a duty to monitor every passage, and absence of enforcement is not permission.
Report suspected abuse to [email protected] without attaching unnecessary private text. Include safe request or order IDs when available. Users may contact support if they believe an automated control or enforcement decision is mistaken.