Skip to main content

Privacy Policy

Last updated: October 24, 2026

What we collect

When you use the Manuscript Readiness Check tool, we collect:

  • Your email address: used so we can follow up with your report and relevant resources. You can unsubscribe at any time.
  • Your target journal name: used to personalize the analysis.
  • Your manuscript file: processed to generate your requested review. Do not upload PHI, a HIPAA limited data set, or directly identifiable patient information.

When you submit a contact form or get a review, we also collect your name and any information you provide in your message.

How we use it

  • Your manuscript is analyzed using the subprocessors listed in our Trust Center and handled under purpose-limited access controls.
  • Your manuscript content is not used to train Manusights models or model-provider models.
  • Your email address is used to send your report results, identify signed-in service activity, and follow up about our review services. It is shared with the service providers listed below for those purposes. We do not sell your email address.
  • We use usage analytics to improve reliability and understand which tools people use. Signed-in connector events use your account email as an identifier; anonymous events use a pseudonymous identifier when the connected app supplies one. Connector analytics include tool names, job identifiers, statuses, and aggregate result counts, but not manuscript text, reference text, or report content.

Data retention

  • Source manuscript files: Scheduled for deletion seven days after the review lifecycle settles, except that unpaid previews keep their source files for the same 30-day window as their preview artifacts so a Full Review can still be started. You can request earlier deletion by emailing team@manusights.com.
  • Generated report artifacts: Free preview artifacts are scheduled for deletion 30 days after the review lifecycle settles. Completed Full Reviews and Dossiers remain available in your account without an automatic expiry, unless you request deletion. This also applies to completed reports from Lab Plan workspaces.
  • Manuscript fingerprints: We keep a short fingerprint of each upload: its title and a compact set of word features from its opening text. It lets us recognize a revised version of the same paper and check public publication databases for whether it was published. We clear the stored title two years after upload.
  • Submission outcomes: If you tell us what happened to a manuscript after its review, such as where you submitted it and the journal's decision, we keep that answer with your account so you can track your submissions, and we use answers only in aggregate to tell future authors what to expect. Reporting a journal's decision can earn a free Full Review. We keep your answers while your account exists, and you can ask us to delete them at any time by emailing team@manusights.com.
  • Operational records: Job status, security, support, billing, and audit records may be kept longer when needed for fraud prevention, accounting, legal obligations, or service reliability. These records are minimized and are not used for model training.
  • Email addresses: Retained until you unsubscribe. You can request deletion at any time by emailing team@manusights.com.
  • Contact form submissions: Retained for up to 2 years for business purposes, then deleted.

Browser storage

We use functional browser storage (cookies and localStorage) for the site to work, not for advertising:

  • Sign-in session: a secure cookie keeps you signed in to your workspace.
  • Returning-customer recognition: a small flag remembers that this browser has used Manusights before, so we can offer you a faster sign-in path instead of starting over.
  • Your draft session: we remember details of your most recent upload (such as your email and target journal) so a return visit can pick up where you left off.

You can clear this at any time from your browser settings. Analytics cookies are separate and governed by the cookie banner.

Third-party services

We use the following services to operate the site:

  • Anthropic (Claude API): processes manuscript text for selected analysis tasks. Anthropic's API terms prohibit using API inputs for model training.
  • OpenAI: may process task context for structured-output fallback and selected review tasks. API inputs are not used to train OpenAI models.
  • OpenRouter and DeepInfra: may process manuscript text, supplied images, and review-task context to complete a Full Review when the Claude and GPT routes fail or return no usable answer. This fallback uses MiMo-V2.6-Pro hosted by DeepInfra, requires zero data retention, and disables data collection for training.
  • Resend: delivers email notifications. Your email address is passed to Resend for delivery purposes.
  • PostHog: records product usage and reliability events, including account email for signed-in connector activity and pseudonymous identifiers where available for anonymous activity.
  • Vercel: hosts the website. Vercel may collect standard server access logs (IP address, browser type, pages visited).
  • Supabase: used for authentication, durable job state, billing metadata, and operational app records.
  • Google Cloud: runs the core review API and worker services in us-east1.
  • Amazon Web Services (AWS): runs review worker services in us-east-1 (Virginia), alongside Google Cloud. Workers there process manuscript text for the review you requested.
  • Stripe: processes payments, receipts, and refunds. Stripe receives billing information, not manuscript content.
  • Public research databases (OpenAlex, Semantic Scholar, Crossref, PubMed, arXiv, Open Library, CRAN): we search them with your paper's title, short search phrases, and the references you cite, to check those references, find related work, and see whether a reviewed paper was later published. They receive search queries, not your manuscript file.
  • Consensus and Paperclip: paper search services. We search them with short queries built from your manuscript, and with the titles of papers you cite, to find related work and check references for Full Reviews and Dossiers. They receive those queries and titles, not your manuscript file.
  • Pangram: for a Dossier, checks your manuscript's writing for signs of AI generation. It receives the opening and closing passages of your manuscript text.
  • Upstash: a short-term cache for reference lookups. A looked-up record is kept for up to 24 hours and a search query for up to 6 hours, so a repeated check does not query the databases again.
  • YouTube (Google): some blog posts embed our own videos with YouTube's player. When one of those pages loads, your browser connects to YouTube, which may set cookies and collect usage data under Google's privacy policy. Pages without a video make no connection to YouTube.
Anthropic

Anthropic Processing Controls

Manusights uses Anthropic for selected review tasks, and its Anthropic API organization operates under a zero data retention (ZDR) agreement with Anthropic. Under ZDR, Anthropic does not store the inputs or outputs of these requests, except where needed to comply with law or to combat misuse or harm, and it keeps the results of its automated safety classifiers to enforce its Usage Policy. Anthropic requires limited retention for its Mythos-class “Covered Models”; Manusights does not use those models for manuscripts. Anthropic API inputs and outputs are not used to train Anthropic's models.

Under this arrangement, manuscript content processed by Anthropic is:

  • Covered by Manusights’s zero data retention agreement with Anthropic
  • Not stored by Anthropic after processing, except as needed to comply with law or combat misuse
  • Not used to train Anthropic models
  • Processed for the requested review task
  • Not used for advertising, resale, or unrelated third-party access

OpenAI API inputs are also not used for training by default. OpenAI states that standard API abuse-monitoring logs may be retained for up to 30 days unless an organization has approved stricter data controls. Manusights is a non-PHI service and does not represent these provider terms as HIPAA compliance.

Questions about this arrangement? Email team@manusights.com.

Connected apps

You can allow another application, such as an AI assistant, to use your Manusights account on your behalf. An app can use your account only after you sign in and approve it on a Manusights consent screen that names the app.

  • Without an account: reference checks and journal AI-policy lookups also work in a connected app without signing in. We then receive only the references or the journal name. For rate limits and usage counts we use a one-way hash of the anonymous identifier the app sends (ChatGPT sends one), or, when an app sends none (Claude sends none), the network address the request comes from.
  • What a connected app can do with your account: check reference lists, look up journals' AI policies, send manuscripts or manuscript text you give it, start previews, and read the reviews on your account. Apps built on our developer API can also open a Stripe checkout page for a paid review; the ChatGPT plugin and the Claude connector cannot.
  • What it cannot do: charge you. Payment always happens on Stripe's own checkout page, which you confirm, and the app never receives your card details.
  • What we share with it: the email address on your account and the reviews belonging to it. Nothing else about you is exposed.
  • Reference checks: the app sends us only the references you ask it to check, not the manuscript itself. We look each one up in Crossref, the DOI registry and Retraction Watch, and, for references without a DOI, in Semantic Scholar, PubMed, arXiv, Open Library and CRAN, and return the results. We keep a record that a check ran and how many references fell into each result, not the references; the lookups themselves are cached for up to 24 hours, as described under Upstash above.
  • Institution licences: if your university or department licenses Manusights, we use the domain of your account email (the part after the @) to apply the licence. We may tell a licensing institution how many accounts on its email domains used Manusights. We never share names, email addresses or anyone's work with it.
  • Removing access: open connected apps in your account and disconnect it. That immediately revokes the app's access, and its existing sessions stop working.

Manuscripts and manuscript text sent through a connected app are handled exactly like manuscripts uploaded on this website, under the retention and processing terms above. The app's own handling of your data is governed by its privacy policy, not ours.

Your rights

You have the right to access, correct, or delete any personal information we hold about you. To exercise these rights, email team@manusights.com and we will respond within 30 days.

Contact

Questions about this policy? Email team@manusights.com.