Legal & Privacy
Terms of Service
These Terms of Service ("Terms") govern your access to and use of the AI chat and data-masking gateway service (the "Service", provided under the brand "Filterit") provided by the operator of filterit (the "Operator", "we", "us", "our").
By creating an account, accessing, or using the Service, you ("you", "User") agree to be bound by these Terms, the Acceptable Use Policy (the "AUP"), the Privacy Policy, the AI Disclosure, the Cookie Notice and, for organisation workspaces, the Data Processing Agreement, each of which is incorporated into these Terms by reference. If you do not agree, do not use the Service.
1. Eligibility and Acceptance
1.1 You must be at least eighteen (18) years of age and have the legal capacity to enter into a binding contract to use the Service. The Service is not directed to, and may not be used by, persons under 18.
1.2 If you use the Service on behalf of an organisation, you represent and warrant that you are authorised to bind that organisation, and "you" refers to both you and that organisation.
1.3 By accepting these Terms you confirm that you meet the eligibility requirements in this Section 1 and that all registration information you provide is accurate and current.
2. Description of the Service
2.1 The Service is an AI chat and data-masking gateway. Its core processing pipeline operates as follows:
(a) a User message (or uploaded document/image) is received;
(b) the Service attempts to detect personal data ("PII") in the content using automated detectors (rule-based validators for identifiers and named-entity models, among them a Greek model trained by the Operator, all run on the Operator's infrastructure; see the AI Disclosure);
(c) detected values are replaced with reversible tokens of the form [NAME_1], [EMAIL_2], etc.;
(d) the masked (pseudonymised) content is transmitted to a third-party large-language-model ("LLM") provider for inference;
(e) the provider's reply is un-masked (tokens replaced with the original values) before being displayed to you.
2.2 The Service is offered through the web application and, by plan, through further surfaces: a Masking API with API keys, an MCP server, editor plugins for VS Code and JetBrains IDEs, a browser extension for Chrome and Edge, a Word add-in, a Windows desktop agent available to signed-in accounts on the Team and Enterprise plans, the Email Assistant (Gmail and Outlook via OAuth, including automatic drafts you review before sending), meeting transcription and summaries, a document review grid, and content-free webhooks. It also offers document and image upload with optical character recognition ("OCR") and a visual-redaction feature that burns opaque boxes over detected personal data and non-text elements in images and PDFs. Dedicated-server, air-gapped, relay and confidential-computing editions are available by pilot arrangement only. Software development kits in our repositories are not published releases.
2.3 Nature of masking (important). The masking performed by the Service is reversible pseudonymisation, not anonymisation. A mapping between each token and its original value is retained in encrypted form (AES-256-GCM) in a secure store (the "Vault"). Under the General Data Protection Regulation ("GDPR") and comparable laws, pseudonymised content remains personal data. We do not represent that the Service anonymises data, nor that "no personal data leaves your server". What is transmitted to third-party AI providers is masked/pseudonymised content (tokens substituted for detected identifiers); the original values are retained by the Operator as described in the Privacy Policy. Token mappings are deleted 30 days after they are created; placeholders in older conversations can then no longer be resolved. Conversation history is kept in masked and encrypted form until you delete it or close your account. Uploaded files are held on temporary storage for at most 30 minutes. Email threads are fetched live and never stored.
2.4 Operator-funded model. The Operator holds the account and API keys with the third-party AI provider (currently Anthropic, PBC), and you consume AI inference under the Operator's account. Consequently, your use of the Service is subject to, and you must comply with, the usage policies of that provider (see Section 5).
2.5 Your explicit choices. The Service offers options that send content to the AI provider without masking: the "RAW" masking mode and the "Original" file mode. They are off by default, marked with a warning, and may be disabled by an organisation's policy. When you choose them, the content you send is not pseudonymised and you alone are responsible for it. Detection also processes the plaintext of your Inputs on the Operator's servers before masking.
3. Accounts and Registration
3.1 To use the Service you register with a valid email address and a password (stored only in hashed form) or sign in with a Google account; you may optionally provide a display name. While registration is by invitation, an invitation code is also required.
3.2 You are responsible for maintaining the confidentiality of your credentials and for all activity under your account. You must notify us promptly at contact@filterit.ai of any unauthorised use.
3.3 We may refuse, suspend, or terminate accounts as set out in Section 12.
3.4 Customer-held key. For individual accounts that sign in with a password, the token mappings and stored content of the account are encrypted under a key derived from your password. The key is unlocked while you are signed in, and during that time the Service decrypts your content in order to serve you. The Service asks you to save a recovery key. If you lose your password and have no recovery key, a password reset leaves your earlier history and mappings permanently unreadable; the Operator cannot recover them. Organisation workspaces and Google sign-in accounts that have not set a password are encrypted under keys held by the Operator instead.
4. User Responsibilities and Lawful Basis for Inputs
4.1 You are solely responsible for the content you submit to the Service ("Inputs") and for ensuring you have all rights and a valid legal basis to do so.
4.2 Third-party and special-category data. Where your Inputs contain personal data of any third party (including special categories under Article 9 GDPR such as health, biometric, or similar data), you represent and warrant that you have a lawful basis and all necessary consents and notices to submit and process that data through the Service and to transmit it (in masked form) to third-party AI providers. As between you and the Operator, you act as controller of any third-party personal data you input, and you appoint the Operator and its sub-processors to process it on your behalf for the purpose of providing the Service.
4.3 You must not submit data that you are not authorised to share. Because PII detection is automated and best-effort (Section 8), you must not rely on the Service to remove personal data you are not entitled to disclose.
4.4 You must not attempt to defeat, circumvent, or degrade the masking, safety, or security mechanisms of the Service, including by prompt injection, by attempting to extract or infer other users' data, or by attempting to reverse-engineer the tokenisation, other than through the options the Service itself offers (Section 2.5). These and other prohibited activities are detailed in the AUP.
5. Flow-Down of Third-Party AI Provider Policies
5.1 The Service currently uses one third-party sub-processor for LLM inference on masked content: Anthropic, PBC (Claude API, United States). We may add, replace or remove providers with notice as provided in Section 12.3; the current list is kept on the sub-processor page. A model running on the Operator's own infrastructure or at an EU provider is in development and not yet available.
5.2 Because you consume AI under the Operator's account, you agree to comply with the applicable usage policies and acceptable-use requirements of the provider, including Anthropic's Usage Policy and the policies of any provider added under Section 5.1, as updated from time to time. You must not use the Service to do anything those policies prohibit.
5.3 A breach of a provider policy is also a breach of these Terms and the AUP, and may result in suspension or termination under Section 12.
6. Intellectual Property and Ownership of Inputs and Outputs
6.1 As between you and the Operator, you retain all rights you hold in your Inputs. You grant the Operator a worldwide, non-exclusive, royalty-free licence to host, process, transmit (in masked form), and display Inputs solely to operate and provide the Service to you.
6.2 Subject to your compliance with these Terms and to the rights of third-party AI providers, the Operator does not claim ownership of the AI-generated responses ("Outputs") returned to you. You are responsible for your use of Outputs. Outputs may be non-unique and similar Outputs may be generated for other users.
6.3 All rights in the Service itself (software, models, interfaces, trademarks, and the "Filterit" name) remain the exclusive property of the Operator and its licensors. No rights are granted except as expressly set out here.
6.4 You grant the Operator the right to use de-identified, aggregated, or technical operational data to maintain, secure, and improve the Service, consistent with the Privacy Policy.
7. Fees, Subscriptions, and Withdrawal
7.1 Some features are offered under paid subscription plans (currently Pro, Team and Enterprise); a Free plan with monthly allowances is available without payment. The current plans, prices, billing periods and plan limits are published on the pricing page and shown again at checkout; those shown at checkout form part of the contract. Prices are shown excluding VAT; VAT is added at checkout according to your country and, for businesses, your VAT number. Enterprise prices are agreed individually. Allowances (AI tokens, masking characters, upload size, request rates) are enforced per plan; when one is exhausted the affected feature pauses until the allowance resets or the plan is upgraded.
7.2 Subscriptions are billed in advance, monthly or annually, and renew automatically for the same period until cancelled. You may cancel at any time from the billing settings; cancellation takes effect at the end of the paid period, the plan stays available until then and no further charge is made. Apart from the right of withdrawal in Section 7.3 and your statutory rights, fees already paid for the current period are not refunded on cancellation. On a Team plan, adding or removing seats is prorated on the next invoice. We may change prices prospectively with notice under Section 12.3; a change applies from your next renewal.
7.3 EU consumer right of withdrawal. If you are a consumer in the EU, you may withdraw from the purchase of a paid subscription within 14 days of the purchase without giving a reason, and we refund the full amount paid within 14 days of receiving your notice, by the payment method you used. We do not ask you to waive this right and starting to use the Service does not end it. To withdraw, email contact@filterit.ai from the address of your account, stating the account email and the date of purchase; you may also use the model withdrawal form of Annex I(B) of Directive 2011/83/EU.
7.4 AI use within the plan and fair use. Use of the AI models through the Service is included in your plan up to the allowance published for the plan on the pricing page (in tokens or as an amount in euro per month) or, where no allowance is published, within fair use: use by people, for the purposes of the plan, in a volume in line with the usual use of customers on the same plan. To check fair use we count the cost of every call to the AI provider in euro, with a price table per model that we may update. The Service chooses the model that answers. We may apply a daily cap as a share of the monthly allowance, offer prepaid credits beyond the allowance and, for automated, resold or abusive use, slow down or suspend the AI feature with notice under Section 12. Included allowances are reviewed at most quarterly and, for the duration of a paid period, only upwards. Masking and the other tools of the Service are not affected by the AI allowance.
8. Limits of Automated PII Detection and Pseudonymisation (No Warranty)
8.1 PII detection is automated and best-effort. The detectors used by the Service are statistical and rule-based and can and will sometimes miss, mislabel, over-detect, or under-detect personal data, including, but not limited to, situations where an optional detector is offline, where text is OCR-derived, where language or formatting is unusual, or where novel identifiers appear.
8.2 The Operator makes no warranty or representation that all personal data in your Inputs will be detected, masked, redacted, or removed before transmission to third-party AI providers, or that the visual-redaction feature will cover every sensitive element. You must independently verify any masking or redaction before relying on it, and you must not submit data you are not authorised to share (Section 4).
8.3 Fail-closed design. By default, if any PII detector is unreachable the affected request is refused rather than sent un-masked. An API client may explicitly request processing with a degraded detector set; the response is then marked as degraded. This is a safety measure, not a guarantee that masking is complete or correct when a request proceeds.
8.4 Nothing in this Section limits any liability that cannot be excluded under mandatory law.
9. Disclaimer of Warranties ("AS IS")
9.1 To the maximum extent permitted by applicable law, the Service, including all Outputs, is provided "AS IS" and "AS AVAILABLE", without warranties of any kind, whether express, implied, or statutory, including any implied warranties of merchantability, fitness for a particular purpose, accuracy, non-infringement, or uninterrupted or error-free operation.
9.2 Outputs are AI-generated and may be inaccurate, incomplete, biased, or fabricated ("hallucinations"). The Service does not provide legal, medical, financial, or other professional advice. See the AI Disclosure. You are responsible for independently verifying Outputs before relying on them.
9.3 Some jurisdictions do not allow the exclusion of certain warranties; in those jurisdictions the exclusions apply to the fullest extent permitted.
10. Limitation of Liability
10.1 To the maximum extent permitted by applicable law, in no event will the Operator, its affiliates, or its suppliers be liable for any indirect, incidental, special, consequential, exemplary, or punitive damages, or for any loss of profits, revenue, data, goodwill, or business, arising out of or relating to the Service, however caused and on any theory of liability.
10.2 Liability cap. The Operator's total aggregate liability arising out of or relating to the Service and these Terms is limited to the fees you paid to the Operator for the Service in the twelve (12) months preceding the event giving rise to the claim, and in any case is not less than one hundred euros (EUR 100).
10.3 Nothing in these Terms limits or excludes liability for intent (δόλος) or gross negligence (βαριά αμέλεια), for death or personal injury, or where the applicable law does not allow a limitation; nor does it affect the statutory rights of consumers that cannot be waived.
11. Indemnity
11.1 You agree to indemnify, defend, and hold harmless the Operator, its affiliates, officers, employees, and agents from and against any claims, liabilities, damages, losses, and expenses (including reasonable legal fees) arising out of or related to: (a) your Inputs; (b) your breach of these Terms, the AUP, or applicable law; (c) your submission of personal data without a lawful basis or required consents; (d) your violation of any third-party AI provider policy; or (e) your misuse of the Service.
12. Suspension, Termination, and Changes
12.1 We may suspend or terminate your access, with or without notice, where you breach these Terms or the AUP, where required by law or by a provider policy, or to protect the Service, other users, or third parties.
12.2 You may stop using and delete your account at any time via account settings. Sections that by their nature should survive termination (including Sections 4, 6, 8 to 11, 13 and 14) survive.
12.3 We may modify these Terms or the Service. For material changes we will give reasonable prior notice (e.g. by email or in-app). Your continued use after the effective date constitutes acceptance; if you do not agree, you must stop using the Service.
13. Governing Law and Jurisdiction
13.1 These Terms are governed by the laws of Greece and applicable European Union law, without regard to conflict-of-laws rules.
13.2 The courts of Athens, Greece have jurisdiction over disputes arising out of or relating to these Terms. If you are a consumer, you keep the mandatory protections of the law of your country of habitual residence, and you may bring proceedings, and may be sued only, before the courts of that country.
14. Miscellaneous
14.1 Entire agreement. These Terms, together with the incorporated documents, constitute the entire agreement between you and the Operator regarding the Service.
14.2 Severability. If any provision is held unenforceable, the remaining provisions continue in full force.
14.3 No waiver. Failure to enforce any provision is not a waiver.
14.4 Assignment. You may not assign these Terms without our consent; we may assign them in connection with a merger, acquisition, or sale of assets.
14.5 Language. These Terms are available in Greek and English; in case of conflict and to the extent permitted by law, the Greek version prevails for Users in Greece.
15. Contact
Questions about these Terms: contact@filterit.ai. Data-protection enquiries: contact@filterit.ai (we have not appointed a data protection officer).