Humanitext OCR

Privacy Policy / Terms of Service

Last updated: August 3, 2026 (consent version 2026-08-03)

Privacy Policy

This Policy sets out how the Humanitext Project ("we", "us" or "our") handles personal information in the OCR service "Humanitext OCR" (the "Service"), provided at https://ocr.humanitext.ai. The Service is a web application that reads characters from PDFs and images uploaded by users, using Google's Gemini API, and produces text and searchable PDFs.

Article 1 (Operator and Contact Point)

  • Operator: the Humanitext Project (Center for Digital Humanities and Social Sciences, Nagoya University, https://dhss.nagoya-u.ac.jp/ ). Its members are Naoya Iwata (Nagoya University / National Institute of Informatics), Ikko Tanaka (J. F. Oberlin University) and Jun Ogawa (The University of Tokyo).
  • Representative (person responsible for personal information): Naoya Iwata
  • Address: Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8601, Japan
  • Contact point for inquiries and for all requests: https://forms.gle/1NTz78iubG9bwrgQA

Article 2 (Information We Collect)

  1. Account information: the identifier (UID), email address and display name obtained when you sign in with your Google account. Authentication is carried out through Firebase Authentication, and we do not hold your password.
  2. Uploaded materials and processed data: originals (PDFs and images), page images, OCR results (text, line coordinates and pre-proofreading backup data) and outputs (ZIP files, combined text, searchable PDFs and bulk ZIP files).
  3. Job history: file names, page counts, dates and times, settings (including the text of prompts) and the email addresses of invited proofreaders.
  4. Usage logs: UID, page count, processing mode, prompt character count (the prompt text itself is not stored), model name, credits consumed, token counts and timestamps.
  5. Credit allowances.

The Service uses no advertising trackers and keeps no records that track browsing behavior. The application does not read the IP address of the requester (it is, however, recorded in the operational logs of the underlying Google Cloud). The Service uses only the cookies (or equivalent technologies) necessary to maintain your sign-in state and for security. We do not use cookies for analytics or advertising. In addition, we use a cookie that stores your display-language preference and browser local storage that stores your theme preference, solely to remember those settings.

Article 3 (Purposes of Use)

  1. Providing the Service (generating page images, running OCR and automatic proofreading, creating searchable PDFs, providing the side-by-side proofreading viewer against the original, generating and delivering outputs, and accepting bulk submission of multiple PDFs and providing bulk ZIP files)
  2. Sign-in authentication and account management
  3. Calculating credits and managing usage allowances, and providing and billing for paid plans
  4. Preventing misuse, and investigating and responding to failures
  5. Compiling usage statistics (page counts, processing modes, credits consumed and the like)
  6. Sending important notices concerning the Service

We do not use images of uploaded materials, OCR results or the text of prompts for analysis aimed at improving the Service or for research.

Article 4 (Transmission to External Services and Subprocessors)

In providing the Service, we entrust the handling of information to the providers listed below. In every case the entrustment is made under our instructions, and we do not permit these companies to use users' information for their own purposes. The only items handled outside Japan are sign-in authentication (Firebase Authentication), the execution of OCR (the Gemini API), and the operational logs that each company retains for its own operations. The country in which processing takes place outside Japan is the United States. The United States is not recognised as having a personal information protection regime equivalent to Japan's; we therefore entrust the handling only after confirming that each company continuously implements protective measures equivalent to those required by Japan's personal information protection law (contractual measures including standard data protection clauses, together with technical measures), and we check the status of that implementation periodically. On request, we will provide information about the implementation of those measures.

SubprocessorRolePlace of processing
Google (Firebase Authentication)Sign-in authenticationUnited States
Google (Gemini API, paid tier)Running OCR and automatic proofreadingUnited States
Google (Cloud Storage)Storing originals, page images, OCR results and outputsJapan (Tokyo region)
Google (Cloud Firestore)Storing account information, job history, usage logs and credit allowancesJapan (Tokyo region)
Google (Cloud Run / Cloud Tasks)Running the application, and scheduling and re-running processingJapan (Tokyo region). Operational logs are handled according to Google's standards
Google (Google Forms)Receiving inquiries and requestsUnited States

What is sent to the Gemini API is the entire image of each page and the full text of the prompt (where automatic proofreading is used, the OCR result from the preceding step is also sent). Because we use the paid tier, the content sent is not used to train models. If you select Batch, processing is carried out after a queue of up to 24 hours and the credits consumed are halved (automatic proofreading is not performed with Batch and is therefore not charged).

We plan to introduce Stripe for payments and Cloudflare for DNS and CDN in the future. When we do, we will revise this Policy in advance and give notice in accordance with Article 11 (Revisions to this Policy). If Stripe is introduced, credit card information will be handled directly by Stripe and will not be retained by us. What we retain is the customer ID, the contracted plan, the number of units and the expiry date.

Article 5 (Retention Periods)

DataRetention period
Uploaded original filesDeleted automatically 30 days after creation, the same as page images and outputs. Uploads that were never turned into a job (including test-OCR-only use) are deleted automatically 48 hours after upload
Copies sent to the Gemini API for batch processing (page images and prompts)Retained by Google for up to 48 hours from receipt, then deleted automatically
Page images and OCR results (text, line coordinates and pre-proofreading backup data)Automatically deleted 30 days after creation
Outputs (ZIP files, combined text, searchable PDFs and bulk ZIP files)Automatically deleted 30 days after creation
Account information, job history and usage logsUntil the account is deleted
Credit allowancesReset on the first day of each month (Japan time). The records themselves are kept until the account is deleted
Infrastructure operational logs (Google Cloud and others)According to each provider's standards (we cannot change these)

A searchable PDF is the uploaded material (a PDF, or page images generated from images) with a transparent text layer overlaid on it, and therefore contains content equivalent to the original. Outputs are deleted after 30 days and cannot be restored thereafter, so please download and save anything you need within that period.

Article 6 (Provision to Third Parties)

We do not provide personal data to third parties without your prior consent, except in the cases below. Provision to the subprocessors listed in the preceding Article, and sharing with proofreaders at your own direction, do not constitute provision to a third party.

  1. Where required by law (including inquiries from investigative authorities and court orders)
  2. Where necessary to protect a person's life, body or property and it is difficult to obtain your consent
  3. Where personal data is transferred in connection with a business succession

Article 7 (Sharing by Inviting Proofreaders)

Users on a paid plan may, for jobs on which searchable PDF is enabled, invite up to 10 proofreaders per job by specifying their email addresses. An invited person can view the page images and OCR results for that job (which contain content equivalent to the original) and can correct and save them line by line. The email addresses of the people you invite are stored as part of the history of that job. Please issue invitations at your own responsibility, understanding that the content of the materials will be visible to the recipient.

Article 8 (Automated Processing by AI)

OCR and automatic proofreading are automated processing by AI, and the output may contain errors or omissions. This processing does not automatically make decisions that produce legal effects or similarly significant impacts on users. Always check the outputs against the original before using them.

Article 9 (Security Measures)

  • All communications are encrypted using TLS.
  • Storage areas and databases are encrypted at rest by the providers' standard functions.
  • Authentication is entrusted to Firebase Authentication, and we do not hold passwords.
  • Access controls are in place so that each user's data can be accessed only by that user and by invited proofreaders.
  • Administrative privileges are limited to the minimum necessary personnel, and access is recorded in the underlying infrastructure logs.
  • All data, including originals, is deleted automatically once the retention periods above have elapsed.
  • Awareness of the external environment: because sign-in authentication and OCR are carried out in the United States, we implement our security measures with an understanding of that country's personal information protection regime.

Article 10 (Requests for Disclosure, Correction or Deletion, and Complaints)

Under applicable law, you may request, with respect to the personal information we hold, notification of the purpose of use, disclosure (including disclosure by electromagnetic record), correction, addition, deletion, suspension of use and suspension of provision to third parties. Please contact us through the inquiry form given in Article 1 (Operator and Contact Point). After verifying your identity, we will respond within one month as a general rule. No fee is charged. We also accept requests made by a legal representative or by an agent you have authorised. In that case we ask for documents confirming their authority.

Withdrawal (deletion of your account) and deletion of individual jobs cannot currently be carried out from the screen. If you make a request through the inquiry form, we will carry out the deletion. Please note that we cannot delete your Google account itself.

Complaints about the handling of personal information may be made to the contact point above and, in Japan, also to the Personal Information Protection Commission.

Article 11 (Revisions to this Policy)

If we revise this Policy, we will give notice of the revised content and of the date on which it takes effect, both on the Service and to your registered email address, at least 30 days before the effective date (except for changes that conform to the general interests of users and minor changes). For changes with a significant impact on users, such as adding a subprocessor, expanding the scope of the information sent or stored, or extending a retention period, we will ask for your consent again.

Terms of Service

Article 1 (Application and Definitions)

  1. These Terms set out the conditions for use of the Service, which we provide at https://ocr.humanitext.ai. The operator and contact details are as set out in Article 1 (Operator and Contact Point) of the Privacy Policy.
  2. "User" means a person who registers an account and uses the Service, and a person who uses the proofreading viewer upon invitation (an "Invitee"). These Terms apply to Invitees as well.
  3. "Content" means the materials uploaded by a User (the "Originals"), the prompts entered, and the page images, OCR results, searchable PDFs, combined text, ZIP files and the like generated from them (the "Outputs").
  4. "Job" means the unit of processing for a single PDF, or for a set of image files (one volume). Bulk submission may cover up to 50 volumes at a time.
  5. A User is deemed to have agreed to these Terms at the point at which the User begins using the Service.

Article 2 (Accounts)

  1. Use of the Service requires signing in with a Google account.
  2. Users shall manage their account at their own responsibility and shall not allow a third party to use it, lend it or transfer it. Acts carried out through an account are deemed to be acts of that User.
  3. Persons under 16 should use the Service with the consent of a parent or legal representative.
  4. Withdrawal (deletion of your account) cannot currently be carried out from the screen. Please make a request through the inquiry form given in Article 1 (Operator and Contact Point) of the Privacy Policy.
  5. We may suspend or delete an account without prior notice where there is a breach of these Terms, acquisition or use of credits by improper means, conduct that causes serious disruption to the operation of the Service, or other reasonable grounds.

Article 3 (Prohibited Conduct)

Users shall not engage in any of the following conduct.

  1. Conduct that violates laws and regulations or public order and morals
  2. Conduct that infringes the copyright, privacy, reputation or other rights of third parties, and uploading materials for which the User does not hold the necessary rights or permissions
  3. Entering content that is discriminatory or hateful, that is intended to harass or defame, or that promotes crime, and using or distributing output obtained thereby
  4. Attempting to reverse engineer the Service or otherwise analyze its internal structure
  5. Excessive access by automated means, circumventing credit limits, and improper use through the creation of multiple accounts
  6. Reselling the Service or the Outputs without our consent, or providing them as a service to third parties
  7. Being, or being involved with, antisocial forces, or engaging in threatening behavior, unreasonable demands or other conduct that obstructs our operations
  8. Any other conduct that interferes with the operation of the Service

Article 4 (Representations and Warranties Regarding Uploaded Materials)

  1. The User represents and warrants that, for all materials it uploads, it holds the rights or permissions necessary to upload them, to render them machine-readable, and to create and use the Outputs (including permission from copyright holders, holding institutions and other rights holders).
  2. Where materials contain personal information, sensitive personal information, trade secrets or other confidential information, the User shall, at its own responsibility, comply with applicable laws and regulations and with the rules of its affiliated institution.
  3. If a dispute arises with a third party in relation to the Content or the use of the Service, the User shall resolve it at its own responsibility and expense. If we suffer damage as a result, the User shall compensate us for that damage.

Article 5 (Ownership of Rights)

  1. Rights in the Originals remain with the User or the original rights holder and are not transferred to us. Rights in the Outputs belong to the User, and we claim no rights in them.
  2. We store, reproduce, convert and transmit Content only to the extent necessary to provide and maintain the Service, respond to failures and prevent misuse. This permission ends upon the lapse of the retention period or the deletion of the account.
  3. Rights in the software and screen design of the Service belong to us or to the rightful rights holders.

Article 6 (Handling of Data and Retention of Outputs)

  1. The handling of Content and of User information, transmission to external services, subprocessors and retention periods are governed by the Privacy Policy.
  2. Outputs are automatically deleted 30 days after creation and cannot be restored thereafter. Please download and save any Outputs you need within that period. We do not provide the Service as a place to store data or as a backup method.
  3. Proofreader invitations may be issued by Users on a paid plan, for jobs on which searchable PDF is enabled, for up to 10 people per job. Invited persons can view the content of that job (which includes content equivalent to the original).

Article 7 (Credits and Usage Allowances)

The processing volume of the Service is calculated in credits. Credits consumed are calculated as "page count x material type weight x automatic proofreading factor x searchable PDF factor".

CategoryFactor
Material type: typeset2
Material type: standard4
Material type: high precision12
Automatic proofreading ONx2
Searchable PDF generationx1.5
Batch processingHalf
  1. The free plan grants 120 credits each month. Paid plans provide 2,400 credits per unit each month (roughly 600 pages with standard OCR).
  2. Credits are a monthly allowance and cannot be carried over to the following month. The allowance is reset on the first day of each month (Japan time). The allowance is granted in full in the month you sign up, but because it resets on the first day of each month, fewer days remain in which to use it.
  3. Credits are not money and cannot be exchanged for cash, refunded or transferred to third parties.
  4. For pages that fail to process, no credits are consumed and any credits are returned automatically.
  5. Where processing completes normally, credits will not be returned and no refund will be made, even if the recognition accuracy falls short of expectations.

Article 8 (Paid Plans and Fees)

This Article applies from the point at which we begin offering paid plans. At present we do not offer paid plans, and the Service may be used only within the free allowance. Even after we begin offering paid plans, no charge will arise unless the User signs up for one.

Prices and contract terms (including the contract period, payment methods, and whether plans renew) are currently under consideration. If and when we begin offering paid plans, the specific prices and terms will be announced on the pricing page and in these Terms beforehand.

  1. Monthly allowances cannot be carried over (Article 7 (Credits and Usage Allowances)).
  2. Fees paid are not refunded, except where required by law.
  3. The handling of information by payment providers is as set out in Article 4 (Transmission to External Services and Subprocessors) of the Privacy Policy.

Article 9 (Changes to, Suspension of, and Termination of the Service)

  1. We may change the content of the Service, and may temporarily suspend all or part of the Service due to maintenance, system failures, outages of external services, natural disasters or other unavoidable circumstances. Except in emergencies, we will endeavor to give advance notice on the Service.
  2. We may terminate all or part of the Service. If we terminate the Service in its entirety while paid plans are being offered, we will give notice on the Service at least three months in advance and will refund the consideration corresponding to the unexpired period, calculated on a pro rata daily basis or by another reasonable method.

Article 10 (Disclaimers and Limitation of Liability)

  1. The Service makes use of a large language model, and its output may contain errors, omissions or unintended generation. We do not warrant recognition accuracy, fitness for any particular purpose, or that the Service will be free from interruption or error. Users shall check the output against the original and verify it at their own responsibility.
  2. Changes to the specifications of, or the discontinuation of, external services on which we rely may result in changes to the Service or in our being unable to provide it.
  3. Except in cases of our willful misconduct or gross negligence, we bear liability only to the extent set out in this Article. Compensation is limited to ordinary damages actually incurred and does not include lost profits or other indirect or special damages.
  4. The cap on compensation is, for Users on a paid plan, the greater of the amount actually paid during the 12 months preceding the event giving rise to the damage and JPY 10,000; and, for Users using the Service free of charge, JPY 10,000.
  5. The provisions of these Terms, including this Article, do not apply to the extent that they are rendered invalid by consumer contract legislation or other mandatory law, and the protection afforded by such law prevails.

Article 11 (Changes to these Terms)

We may change these Terms where the change conforms to the general interests of Users, or where it is reasonable in light of the necessity for the change and the appropriateness of the changed content. If we make a change, we will give notice of the changed content and of the date on which it takes effect, both on the Service and to your registered email address, at least 30 days before the effective date (except for changes that conform to the general interests of Users and minor changes). For changes with a significant impact on Users, we will ask for your consent again.

Article 12 (Governing Law and Jurisdiction)

  1. These Terms are governed by the laws of Japan.
  2. The Nagoya District Court shall be the exclusive court of first instance by agreement for any dispute arising between us and a User in relation to the Service or these Terms.
  3. Users who are consumers enjoy the protection of those consumer protection provisions of the law of their country of residence that cannot be excluded by agreement. Where such law permits proceedings to be brought before the courts of their place of residence, that right is not impaired.

Article 13 (Governing Language and Contact)

The Japanese text of these Terms and of the Privacy Policy is the authoritative version. Versions in English or other languages are translations provided for reference only, and in the event of any discrepancy in interpretation, the Japanese version prevails. Please direct inquiries, complaints and all requests to the inquiry form at https://forms.gle/1NTz78iubG9bwrgQA.

These Terms and the Privacy Policy were established and took effect on August 1, 2026, and were last revised on August 3, 2026 (version: 2026-08-03).