Privacy policy

Zetesis provides scientific evaluation of claims through the service at api.zetesis.science. This policy explains what we collect, who processes it, and how we use it. Last updated 1 September 2026.

What we collect

• Content you submit (the claim, any context, and any documents) is processed to produce your evaluation.
• Account details: the name, email, organization, and information you provide when you request access, and your access token.
• Usage metadata: the engine-routed domain of a claim, the evaluation depth, counts, timestamps, your IP address, and the client name your software sends in its User-Agent header. We record these for requests to the API and to the MCP server alike, including the tools that run no model. We use them to rate-limit the free tier, to keep the service running, and to understand which clients people connect from. We do not log the text of your claim in our usage records.

How we use it

• To produce and deliver evaluations. A full dossier is stored so that you, and anyone you share its link with, can retrieve it. A free screen is processed and returned; only usage metadata is kept.
• To produce an evaluation, non-identifying search queries are sent to public scientific sources (Europe PMC, ClinicalTrials.gov, openFDA, NIH RePORTER, SEC EDGAR).
• Attestation verification is public by design: an attestation can be re-checked by anyone, without an account.

Who else processes your content

Zetesis is built on a large language model, so you should know exactly where your content goes.

• Anthropic, PBC is our sub-processor. The content you submit (the claim, any context, and any documents) is transmitted to the Anthropic API so the model can scope and grade it. This applies to the free screen and to a full dossier alike. Anthropic's handling of that content is governed by its own terms and privacy policy, at anthropic.com/legal/privacy.
• Under Anthropic's commercial API terms, content submitted through the API is not used to train its models. Anthropic retains API inputs and outputs for a limited period for trust and safety purposes; see its policy for the current window. We have not arranged a zero-retention configuration, so assume that window applies.
• Our own infrastructure runs on Amazon Web Services in the eu-central-1 region.
• The two Zetesis tools that only retrieve evidence, zetesis_scope and zetesis_evidence, do not call a model at all. Content you send to those is used to build public-source search queries and is not transmitted to Anthropic.

If you are working under an NDA or with confidential material, take this section as the operative fact: a third party processes your text. Contact us before submitting anything you cannot send to a model provider.

What we do not do

• We do not sell your data.
• We do not use your submitted content to train models, and our model provider does not train on it either, as above.
• We do not share your content with anyone beyond the sub-processor and the public sources named above.

Retention and your choices

• Dossiers are retained until you ask us to delete them. Access records and account details are retained to manage your access.
• Deleting your data from Zetesis does not by itself clear any copy held by our sub-processor under its own retention window.
• To request access to, or deletion of, your data, email avidan.r@zetesis.science.

Security and contact

Content is transmitted over TLS and access is token-gated. Questions, or to report a security issue: avidan.r@zetesis.science.

Back