Skip to content
Ask Tone

Privacy Policy

Last updated November 2, 2026

This policy explains what information Ask Tone collects, how we use it, and the choices you have. We have tried to write it in plain language rather than legalese.

The short version

  • We collect the minimum we need to run your account, take payment, and operate the service.
  • We don’t read your conversations, and we don’t use your content to train any of our own systems.
  • To answer your questions, your prompts are sent to third-party AI providers we don’t own. Each has its own policy that governs what it does with that data, and we explain this below.
  • Ask Tone only carries what you said in one chat into another while memory across chats is switched on. New accounts start with it off, and Settings → Memory shows whether it is on for yours.
  • You can access, export, or delete your data at any time.

Who we are

Ask Tone is a service that lets you put a question to several large language models (LLMs) at once and receive their answers side by side or reconciled into one. Ask Tone is operated by Ask Tone Ltd, a company registered in England and Wales under company number 17322527, whose registered office is at 20 Wenlock Road, London N1 7GU (“Ask Tone”, “we”, “us”). For the purposes of UK and EU data protection law, we are the data controller of the personal data described in this policy.

This policy applies to the Ask Tone website at asktone.ai and the Ask Tone application. If you have any questions, contact us at privacy@asktone.ai.

Information we collect

Account information

When you create an account we collect your email address, and your name if you add one. If you sign in with Google, we receive basic profile information from Google, including your email address and your name, and we use that name as your display name. You can change your name in Settings, or ask us to, at any time. It is shown in your account, to the other members of any team workspace you belong to, and in team invitations you send. We store authentication data needed to keep your account secure, including passkey credentials and password hashes. We never store your password in plain text.

Payment information

Payments are handled by our payment processor, Stripe. We do not receive or store your full card number. We keep records of your plan, transactions, and credit balance so we can provide and bill the service.

Free trial

If you start a free trial, we record when it started and ends and the usage allowance it granted. To limit trials to one per person, we also store a one-way hash (a “fingerprint”) of your email address in a normalised form, never the address itself. The fingerprint cannot be turned back into your email address and is used for nothing else. It is kept even after your account is deleted, so the same address cannot be used to start another trial; see “How long we keep data” below.

Content you provide

This includes the prompts you submit, the conversations you have, any files or text you add to them, and any images you generate. How this content is handled is described under “Your conversations” below.

Usage and technical information

We collect limited technical data needed to operate and secure the service, such as log data and approximate location derived from your IP address. We also count visits to our public pages (our website, the developer documentation, and the sign-in and sign-up pages) without using cookies: our hosting provider records which page was viewed, which site sent the visit, any campaign tags in the link, and your country, browser and device type, and reports these to us only as totals. This never runs inside your account or your conversations. With your consent, we also collect product analytics to understand how the service is used; on our website it runs on those same public pages only. You control product analytics on the website through our cookie banner, and you can read more in our Cookies Policy. Our servers also send our analytics provider operating counts, for example how many answers were generated and how long they took. They carry no name, email address or account identifier and none of your content, and they are not used to build a profile of you.

How we use it

We use the information above to: provide and maintain the service; process payments and manage your subscription and credits; send you essential service messages (such as receipts, security notices, and account updates) and the other emails described under “Emails we send”; provide support; keep the service secure and prevent abuse; comply with our legal obligations; count visits to our public pages, so we know how people find us; and, where you have consented, understand and improve the product. We do not sell your personal data, and we do not use your conversations to advertise to you.

Emails we send

Some emails are part of running your account and always arrive: sign-in and verification emails, receipts and billing notices, security notices, the welcome email when you buy a plan, team invitations, the confirmation when you start a free trial, the warning before we delete an inactive account, and the confirmation when you delete yours.

While you are on a free trial, we may also send you:

  • Trial reminders: a few days before your trial ends, the day before it ends, and when it has ended. These follow the Account & security switch under Notifications in Settings.
  • Tips: at most two short emails early in your trial on getting more from Ask Tone. New accounts receive them unless you untick the box on the screen where you choose a plan, and they follow the Tips switch in the same place.

Each of these emails says how to turn it off. Turning them off does not affect the emails that always arrive.

Your conversations

We treat the content of your conversations as private to you.

  • We encrypt your conversation content, including any images you generate, with a key unique to your account, so it is not readable in our database or our file storage as ordinary text or images.
  • We don’t read your chats. Our staff do not access the content of your conversations in the ordinary course of running the service. Limited access may occur only where strictly necessary, for example to investigate a specific abuse report or where we are legally required to, and such access is restricted and logged.
  • We don’t train on your content. We do not use your prompts, conversations, or outputs to train, fine-tune, or improve any AI model of our own.

There is one important exception you should understand, which is that producing an answer requires sending your prompt to a third party. We explain that next.

Memory

Ask Tone can carry context between messages so you don’t have to repeat yourself. There are three kinds, and they differ in how far the information travels.

  • Your brief. A short note you write about yourself and your work, in Settings → Memory, along with an optional note about your coding preferences. You control the text. It is encrypted with your account key and included in your prompts.
  • Conversation summaries. When a conversation gets long, we condense the earlier part of it so the model can still follow the thread. A summary is made from that conversation and is only ever used in that same conversation. It is encrypted with your account key.
  • Memory across chats. While it is switched on, Ask Tone notes durable facts from your conversations (for example a preference for short answers, or the language you work in) and includes them in later chats, including new ones. This is the only feature that carries information out of the conversation it came from, which is why it is something you switch on rather than something that always runs. New accounts start with it off. Settings → Memory shows whether it is on for your account, and turns it off whenever you want.

Whenever memory across chats is on, Settings → Memory shows you everything it has noted. You can edit any entry, pause the feature, delete individual entries, or clear them all. Turning it off stops both the noting and the use of what was already noted, from your very next message. Entries are encrypted with your account key and, like your conversations, they are included in the prompt sent to the AI provider that answers you, which is how they have any effect at all. We do not use them for advertising, profiling, or training, and deleting your account destroys them with everything else (see “How long we keep data”).

Third-party AI providers

Ask Tone does not build, own, or operate the AI models that answer your questions. To generate the responses you see, we transmit the content of your prompt (and, where relevant, the surrounding conversation) to third-party AI providers. Most of that traffic goes through two routing services we use, OpenRouter and Groq, which pass your request to the underlying model provider (for example, the companies behind the models offered in the app). Some models we reach directly rather than through a router (currently Perplexity), in which case your prompt goes to that company and to no intermediary.

Image generation works the same way: your prompt is sent through OpenRouter to the image model you chose, solely to generate the picture. When you edit a picture you have made, a copy of that picture, reduced in size, is sent with your prompt to the image model, solely to make the new version. Before a generated image is stored or shown to you, it is checked by an automated safety classifier operated by OpenAI. The image is sent to that classifier for the check and for nothing else. Generated images are then stored encrypted, like the rest of your conversations.

Files and images you attach. A document you attach (a text file, a PDF, or a Word, Excel or PowerPoint file) is read as text on our servers, and that text, or the parts of it most relevant to your question, is sent with your message to the models on your board. For an image you attach, we first make a copy for the models: its first frame, reduced in size, without the location and camera details the file may carry. That copy is checked once by OpenAI’s automated safety classifier before we keep anything, and is sent to that classifier for the check and for nothing else. The copy is then sent with your messages only to the models on your board that can see images. What we keep, and what you download from your Collection, is a cleaned copy of the image at its full size: the picture alone, without the location and camera details or anything else the file carried, its first frame if it was an animation, and sometimes in another format (a photo kept as a JPEG, say). Pictures inside a document you attach are checked the same way before we keep the document. If one of them is flagged or cannot be checked, or the document holds something we cannot look inside (a file attached within it, say), we keep the document’s text and not the file. Like your conversations, attached files and images are stored encrypted with your account key and are never used to train any model.

Every message and image prompt is also checked by an automated safety classifier before it is answered. That check runs through OpenRouter, or through Groq when OpenRouter cannot answer. Only when neither can is the text checked by OpenAI’s moderation classifier instead. Whichever runs the check, your text is sent to it for the check and for nothing else.

Web search. When you turn on web search for a message, the text of that message is sent to a web search provider to find current pages that may help answer it. We use one of two: Exa, reached through OpenRouter, or Perplexity’s search service, reached directly. Which one runs is a service-wide setting rather than something you choose, and if the one we try cannot answer (during an outage, for instance), that single message is sent to the other instead, so a message may reach both in the course of one search. Only that message is sent for searching (not your earlier conversation, your files, or your memory), it carries nothing that identifies you, and it is sent only when you choose to search. The pages found are given to the models alongside your question, and the ones the models were shown are listed under the answer and stored encrypted with your conversation, like the rest of it. Web search is off unless you switch it on, and it is not available on a free trial.

Models that search the web themselves. Some models can run their own web search as part of answering, rather than relying on the search described above. Where a model works that way, choosing it means your message goes to that model’s provider as usual and that provider searches the web with it. Because the searching is part of how the model works, this happens whether or not you switched our own web search on. Those models return the pages they used, and we list them under the answer in the same way. The app tells you when this is the case: with such a model selected, the web-search control reads “included” and adds no charge, because the search is the model’s rather than ours.

Dictation. When you press the microphone in the composer, the audio is recorded on your device and sent to a speech-to-text provider to be turned into text, which lands in the message box for you to read and edit before you send it. We do not keep the recording. It is never written to our database or our file storage: it exists in memory for the moment it takes to transcribe, and is discarded. The transcript itself is not stored either until you choose to send the message, at which point it is your message and is encrypted with the rest of your conversation. Nothing is recorded unless you press the microphone, and your browser or phone will ask your permission the first time.

Reading answers aloud. The speaker under an answer uses the text-to-speech built into your own device. The answer is not sent anywhere to be spoken, and no audio is produced by us or by any provider.

Because we do not control these providers or the models they run:

• Each provider processes your prompt under its own privacy policy and terms, which are separate from this one, which we cannot change, and which govern what that provider does with the data we send it. A provider’s policy may permit things this policy does not, may change over time, and in the event of any conflict will govern that provider’s handling of your data rather than this policy.

• OpenRouter and Groq both state that they do not use data submitted through their paid APIs to train their own models, and we configure our use of them to avoid routing to providers that would train on your prompts. However, these practices are set and changed by the providers, not by us, and we cannot guarantee them.

• Some providers temporarily retain prompts and outputs, sometimes for up to 30 days, to monitor for abuse, troubleshoot, or meet legal obligations, even where they do not train on them.

• To make the later messages in a conversation cheaper to process, we ask the provider of some models (currently Anthropic, for its Claude models, reached through OpenRouter) to keep a working copy of the unchanged earlier part of that conversation ready for a short time (about five minutes) after each message, so that your next message can reuse it rather than being processed from scratch. Anthropic describes this copy as held in memory only, deleted when that time is up, and not stored or used for training. It is not a way for anyone to read your conversation: reusing it requires sending the identical text again.

• Where a provider publishes a commitment about one of its services, such as not retaining what is sent to it or not training on it, we do not assume that it extends to that provider’s other services. Perplexity, for example, offers both a service that answers questions and a separate search service, and we use both; a statement about one is not a statement about the other. Read each provider’s own policy for what applies to what.

In short: our commitment not to read your conversations or train our own systems on your content applies to Ask Tone. It cannot bind the independent AI providers your prompts must be sent to in order to produce an answer. If a particular prompt is sensitive, please bear in mind that it will be processed by one of these third parties. You can review their policies at the links below, and the company behind each underlying model has its own policy in addition:

Service providers

We rely on a small number of carefully chosen providers to run Ask Tone, for example to host our application and database and count visits to our public pages, process payments, send email, and monitor the reliability of the service without capturing the content of your conversations. We also use a provider for product analytics: on our website only where you have consented, and from our servers for the operating counts described above. These providers act as our processors: they handle personal data on our instructions, only to the extent needed to provide their service to us, and under written agreements. The third-party AI providers your prompts are sent to in order to generate answers are described separately above, and we keep a current list of the providers we use that we can share on request at privacy@asktone.ai.

Cookies

We use a small set of cookies and similar technologies, and we only set non-essential ones (such as analytics cookies) with your consent. Counting visits to our public pages uses no cookies and stores nothing on your device. Full details are in our Cookies Policy.

How long we keep data

We keep your account information and conversations, including any images you generated and anything memory has noted, for as long as your account is active. You can delete individual memory entries, or clear them all, at any time in Settings → Memory. When you delete a conversation or your account, it becomes inaccessible immediately and is then permanently removed from our systems shortly afterwards, after a short recovery window. That includes any images the conversation contained. When you delete your whole account, we also destroy its encryption key, so any remaining copy of your content cannot be read. We may also delete an account that has been inactive for a long time, and we email you before we do: after about three months without activity if it has never had a paid plan of its own (including one whose free trial has ended), or after about 12 months if its plan has ended. We keep certain records (such as transaction and tax records) for as long as the law requires. One deliberate exception: the hashed free-trial fingerprint described under “Information we collect” is retained after your account is deleted, because its only purpose is to stop a second trial being started with the same address. It is a one-way hash, holds no other information about you, and is never combined with any other data.

Your rights

Depending on where you live, you have rights over your personal data, including the right to access it, correct it, delete it, receive a copy in a portable format, object to or restrict certain processing, and withdraw consent. You can exercise many of these directly in your account settings, or by contacting privacy@asktone.ai. If you are in the UK or EU and believe we have not handled your data properly, you have the right to complain to your data protection authority (in the UK, the Information Commissioner’s Office).

International transfers

Some of our providers, including our AI routing providers, are based in or process data in the United States and other countries. Where we transfer personal data outside the UK or European Economic Area, we rely on appropriate safeguards (such as the UK International Data Transfer Agreement or the EU Standard Contractual Clauses) to protect it.

Children

Ask Tone is not intended for children. You must be at least 18 years old (or the age of majority where you live) to use it, and we do not knowingly collect personal data from children. If you believe a child has provided us with personal data, please contact us and we will delete it.

Security

We take a range of measures to protect your account and data, including per-account encryption of conversation content, encryption in transit, strong authentication options, and strict internal access controls. You can read more on our Security page. No service can be perfectly secure, but we work hard to protect your information.

Changes

We may update this policy from time to time. When we do, we will change the “last updated” date above, and for material changes we will provide additional notice (for example, by email or an in-app notice). Your continued use of Ask Tone after a change means you accept the updated policy.

Contact

For any privacy question or to exercise your rights, contact us at privacy@asktone.ai, or write to Ask Tone Ltd, 20 Wenlock Road, London N1 7GU.