Public Beta operations
We are currently operating in Public Beta. During this period, select execution details (including query text) are temporarily logged, tracked, and reviewed strictly for debugging and system reliability work. We plan to eliminate content logging after beta.
No model training
We do not use your drafts to train Slop Sentry systems, and we do not use them to train the third-party models that process your requests. Draft text is sent to model providers only to run the checks and edits you ask for.
Models we use
Slop Sentry uses more than one model so each stage can use the stack that produces the cleanest slop-free result for your content. Exact pattern matches run in code. Detection and judgment passes use Anthropic Claude. Rewrites and edits use OpenAI models that do not apply Anthropic watermarking. Related Cloudflare infrastructure may also run specialized detection models.
Encrypted in transit
Your text travels over TLS between your client, Slop Sentry, and the model providers that process a request. Diagnostic logs collected during beta are access-controlled and are not published.
Minimal logging
Usage metadata for tracking and billing is our standard. Beyond the temporary public-beta debugging logs, we minimize how long draft text is kept.
Third-party processing
Running Slop Sentry means your draft is processed by Anthropic, OpenAI, and Cloudflare for that request. Do not send material that your policies prohibit third-party AI services from processing.