Join session Instructor sign-in

Policies

Privacy Policy

Version 2026-10-25 · Effective 25 October 2026

TutorTalks is operated by Arclight Innovation LLC. An instructor sets up a spoken practice conversation with an AI avatar, participants join it with a code in their browser, and the instructor reviews what happened. This policy explains what personal information we collect, why we use it, who receives it, how long we keep it, and the choices and rights you have.

If you joined a session as a participant

The organization or instructor that invited you decides how your session records are used, and requests to see, correct or delete them go to them first. Sections 2 and 11 explain how that works. If your session is run by a school or college, the student data and FERPA notice applies as well.

1. The short version

  • Who we are. Arclight Innovation LLC runs TutorTalks. Write to studio@arclightxr.com about anything in this policy.
  • Two roles. For account holders, billing contacts and website visitors, we decide how information is used. For what participants say and produce in a session, we act for the organization or instructor that runs it, on their instructions.
  • What we collect. Account and billing details, what instructors build, and, from participants, an email address, an optional name, their voice, a transcript and visit details, plus AI-written insights when an organization uses that add-on.
  • AI providers process conversations. Speech and text go to AI providers (OpenAI by default, Google Gemini for the AI Session Designer, HeyGen for the avatar) to transcribe, reply, speak and animate. The name a participant types when joining, and their pronoun answer, are part of what the language model receives. The email addresses we hold are not.
  • No selling, no ads, no training by us. We do not sell or share personal information, show ads, or train AI models on customer or participant content. We use providers’ business services and have opted out of provider training wherever an opt-out exists.
  • Limited analytics. Google Analytics runs on four pages: /pricing, /instructor, /studio and /redeem. It does not run on the pages participants use.
  • Set retention periods. Recordings and transcripts follow the organization’s settings (90 and 365 days by default). Proof that you accepted our Terms and this policy is kept for the life of the account plus six years.
  • Your rights. Account holders can ask us for a copy, a correction or deletion by email, and we respond within 30 days. Participants go to the organization or instructor that invited them, and we help them respond.
  • Adults. Account holders must be 18 or older. Participants must be 18 or older unless we have agreed otherwise in writing.

2. Who we are and our two roles

TutorTalks (tutortalks.io) is operated by Arclight Innovation LLC (“Arclight”, “we”, “us”). Email is how to reach us about this policy, including privacy requests: studio@arclightxr.com.

This policy covers our website, the instructor dashboard and studio, the session pages participants use (including sessions embedded in other websites), and our billing. It does not cover websites that embed a session, Google’s sign-in pages, Stripe’s checkout pages, or what a customer does with records after taking a copy out of TutorTalks. Those are governed by their owners’ policies.

Where we decide

We are responsible for (in some laws’ terms, the “controller” of) personal information about people who visit our website; account holders, organization owners, administrators, members and billing contacts; people who receive a credit link from us or write to us; and the records that show someone accepted our Terms and this policy.

Where we act for our customers

Our customer is whoever holds the TutorTalks account: a college or university, a company, a training provider, or an individual instructor. When a customer runs sessions, the participants’ details, voice, recordings, transcripts, visit details and any AI-written insights about them (“participant records”, which our Terms call Session Records) are the customer’s. We process participant records only to provide TutorTalks to that customer, on its instructions and under its settings, such as whether audio is recorded and how long records are kept. In legal terms we act as the customer’s processor or service provider. We act in the same way for any personal information in the content a customer’s members put into TutorTalks, such as knowledge-base documents, uploaded images and approved-email lists (“Customer Content” in our Terms).

We do not use participant records for our own purposes, and we disclose them only to the customer, to the providers that help us run the service (section 6), or where the law requires. The customer is responsible for telling participants about its sessions and for any consent the law requires of it, for example to record a conversation.

Schools and student records

When the customer is a school, college or university subject to the Family Educational Rights and Privacy Act (FERPA), participant records are education records. We act as a school official with a legitimate educational interest, under the institution’s direct control, and the student data and FERPA notice explains how that works.

TutorTalks is designed for postsecondary, professional, corporate and medical training. It is not designed for primary or secondary (K-12) schools. A K-12 school must contact us first and may use TutorTalks only under a written agreement with us.

3. What we collect

What we collect depends on how you use TutorTalks. The tables below set it out by whom it is about and where it comes from. We do not ask for government identification numbers (other than an optional business tax ID at checkout) or collect precise location, and we do not record video of anyone: the avatar’s video streams to the participant’s browser and is never recorded.

Website visitors

InformationWhat it includesWhere it comes from
Server logsIP address, date and time, the address of the page requested, including any code or search text in it (for example a session’s join code, a credit-link code, or a name or email address an instructor types into the Students search), the referring page, and the browser’s description of itself (its user-agent string).Your browser, automatically.
AnalyticsOn /pricing, /instructor, /studio and /redeem only: pages viewed, approximate location derived from your IP address, browser and device details, and an identifier stored in your browser. Section 8 explains how to opt out.Google Analytics, through cookies and similar browser storage.
Error reportsOn the instructor, studio and session pages, when something fails: technical details of the error and the page address, with query strings removed and identifiers masked. Email addresses, sign-in tokens and keys are removed before a report is sent.Your browser, automatically.

Account holders and organization members

InformationWhat it includesWhere it comes from
IdentityName, email address and whether it is verified, profile photo link, sign-in method (Google, or an email address and password) and a user ID. Firebase Authentication stores passwords; our servers do not.You, through Google sign-in or the sign-in form, via Firebase Authentication.
ProfileYour default organization and when you last signed in.Our systems.
OrganizationOrganization name and subject area, billing and contact email, website, owner, members and their roles, and invitations, including the invited person’s email address.You and others in your organization.
What you createSessions and their personas, prompts, guardrails, topics, briefings and settings; approved-email lists; the websites allowed to embed a session; knowledge-base documents (we keep the extracted text and a numeric index of it, not the original file); avatar images and videos you upload; and backdrops.You.
AI Session Designer and AI editYour brief and any files you attach. Attachments are read to produce a draft and are not stored unless you keep the draft.You.
Your voice in previewsWhen you try a session in the studio or a preview, your voice and the conversation are processed the same way as a participant’s, but are not recorded or stored.You.
Plan, usage and creditsPlan, add-ons, subscription status and renewal date, credit and AI-credit balances, minutes and sessions used, and our cost of providing them.Our systems and Stripe.
Security and audit recordsWho did what and when, with the IP address and a request ID: for example viewing, playing, downloading, exporting or deleting a participant record, or changing a role, a retention period or a billing setting. Sign-in attempts are recorded in our server logs with the IP address.Our systems, automatically.
Acceptance recordsThe record made when you accept our Terms and this policy before a payment. Section 7 lists its contents.You and our systems.
CommunicationsWhat you write to us and our replies, and notes we keep about your account to support it.You and us.

Billing contacts

InformationWhat it includesWhere it comes from
Purchase detailsName, email address, billing address (our checkout requires it) and, if you give them, your organization’s name and tax ID.You, on Stripe’s checkout page.
Payment detailsCard or bank details, or a wallet such as Link, Apple Pay or Google Pay. These go to Stripe and never reach our servers.You, directly to Stripe.
Payment recordsCustomer and subscription IDs, invoices, payment status, refunds and disputes.Stripe.

Participants (processed for our customers)

InformationWhat it includesWhere it comes from
Join detailsThe join code; an email address, which is required; a name, which is optional by default and recorded as “Anonymous” when left blank; which version of the session notice the participant acknowledged and when; and how many times they have taken the session.The participant, on the join form.
Pronoun answerAn optional answer that lets the tutor address the participant correctly in languages with grammatical gender. It stays in the browser tab, is included in what the language model receives during the session, and is never stored on our servers or shown to the instructor.The participant.
VoiceSpeech captured by the microphone while the session runs (in hands-free mode the microphone reopens by itself when the tutor finishes speaking), streamed to our server for transcription and the conversation. It is recorded only when recording is on for the session and allowed by the organization, and a recording holds both the participant’s and the tutor’s audio.The participant’s microphone.
TranscriptEvery turn of the conversation, with timestamps, the language, and whether a guardrail blocked a turn.The conversation.
Visit detailsStart and end times, number of turns, speaking time and minutes used.Our systems.
AI-written insightsWhen the organization uses the Student insights add-on: a report on the participant’s session for the instructor and, if the instructor releases it, a copy written for the participant.The language-model provider, from the transcript.
Device and connectionIP address and browser details, used to run the connection and to limit repeated join attempts. The avatar provider and the networks that deliver our page software also see them (section 6).The participant’s browser.
Approved-email listsEmail addresses or email domains an instructor allows to join a session. A person can be on a list without ever joining.The instructor.
In-session notesCoaching and conversation notes the tutor uses during a session. They are held only in our server’s memory while the session runs, are never saved with the session’s records, and are never sent to participants (when instructors preview their own session, the coaching notes reach their own browser). The one trace kept is a line in our server log, held for 30 days, that records the coach’s one-word reading of how the participant is doing (struggling, steady, confident or disengaged) against a shortened visit ID, not a name.The language-model provider.

People who receive a credit link

InformationWhat it includesWhere it comes from
Credit grantsEmail address, number of credits, the reason, who granted them, and whether the link has been used.Arclight, when we send the link.

Please do not put sensitive information, such as real patient records, government ID numbers or payment card numbers, into sessions, knowledge bases or conversations. TutorTalks is not designed to hold it, unless we have agreed otherwise in writing.

4. How we use it

We use personal information only for the purposes below. The last column gives the legal basis we rely on under the EU and UK General Data Protection Regulation (section 13 explains those terms). For participant records, and for personal information in Customer Content, the legal basis is the customer’s, and we act on its instructions.

PurposeInformation usedLegal basis (EEA and UK)
Providing TutorTalks: accounts, sign-in, sessions, the live conversation, recordings, transcripts and reportsIdentity, organization, what you create, participant recordsPerformance of our contract with you. For members, administrators and billing contacts of a customer’s organization who are not themselves party to our contract: our and the customer’s legitimate interests in providing the service the customer has bought. For participant records and personal information in Customer Content, the customer’s instructions.
Billing, metering and creditsPlan, usage and credits; purchase and payment recordsContract; legal obligation, for tax and accounting records.
Keeping proof that you accepted our Terms and this policyAcceptance records (section 7)Legitimate interests in proving our agreement and establishing, exercising or defending legal claims; legal obligation where a law requires us to keep proof of consent to automatic renewal.
Security: preventing abuse and fraud, enforcing limits, investigating incidents and fixing errorsServer logs, audit records, error reports, IP addressesLegitimate interests in a secure, working service.
Understanding how our pages are usedGoogle Analytics on four pages; usage figuresLegitimate interests in improving our website and product. You can object (section 8).
Support and service messages: answering you and telling you about changes to the service, these documents or your accountIdentity, communicationsContract; legitimate interests.
Improving TutorTalksUsage and cost figures, error reports, feedback you sendLegitimate interests. We do not review participant conversations to improve the product unless a customer asks us to help with a specific session.
Complying with law and protecting rights, including enforcing our TermsWhat a request or matter requiresLegal obligation; legitimate interests.
A merger, acquisition or sale of our businessInformation that transfers with the businessLegitimate interests.

What we never do

We do not sell personal information or share it for cross-context behavioural advertising. We do not show advertising in TutorTalks or use personal information for targeted advertising. We do not use participant records or customer content to train or fine-tune AI models, ours or anyone else’s. We do not build profiles of participants for any purpose other than the customer’s own use of TutorTalks, and we do not use anyone’s voice to identify them.

5. AI processing

A spoken conversation with an AI avatar takes several AI services: one turns speech into text, a language model writes the tutor’s replies, another turns replies back into speech, and another animates the avatar. These services are run by the providers named on our Subprocessors page. We choose the provider for each task and may change it, but only to a provider listed on that page.

TaskProvider todayWhat it receives
The tutor’s repliesOpenAI (Google Gemini can be selected)The session’s setup written by the instructor, the participant’s typed name and pronoun answer, the conversation so far, excerpts from the session’s knowledge base, and in-session notes.
Speech to textOpenAI (ElevenLabs can be selected)The participant’s speech.
Text to speechOpenAI (ElevenLabs can be selected)The text of the tutor’s reply.
Speech to speech, when an instructor turns it on for a sessionOpenAIThe participant’s speech and the conversation. The model hears the speech directly and answers aloud.
Avatar videoHeyGen (LiveAvatar)The tutor’s reply audio and the chosen avatar. The participant’s browser connects directly to HeyGen’s media servers to receive the video, so HeyGen also receives the participant’s IP address and browser details. The participant’s microphone is never sent to HeyGen.
Knowledge searchOpenAI (follows the reply provider)Knowledge-base text, and what the participant says, converted into a numeric form used to find relevant excerpts.
Coaching notes, conversation notes and Student insightsThe reply provider (OpenAI) by defaultThe conversation and, for Student insights, the participant’s typed name.
AI Session Designer and AI editGoogle Gemini (OpenAI can be selected)The instructor’s brief, attached files and the session being edited.

We never put an account holder’s or participant’s email address into a request to an AI provider, and approved-email lists are never shown to the AI. An email address reaches a provider only if someone says or types it in a conversation, or includes it in a brief or a document.

No training on your content

Arclight does not train, fine-tune or otherwise improve AI models with customer content or participant records. We use our providers’ business and API services, under terms that limit how they may use what we send, and we have opted out of provider training wherever a provider offers an opt-out. Providers may keep data for a limited time to detect abuse or as their terms require; OpenAI, for example, may keep API inputs and outputs for up to 30 days for abuse monitoring. Where a provider publishes how long it keeps API data, the Subprocessors page says so.

Voice and biometrics

We do not create voiceprints or speaker profiles, and we never use a voice to identify or authenticate anyone. Voice is used only to transcribe and hold the conversation and, where recording is on, to keep the recording for the customer. Where an instructor turns on speech to speech, the model hears the participant’s voice directly and responds to how they sound as well as to what they say (slowing down for someone who sounds unsure, for example), and it can judge speaking criteria such as pronunciation, fluency or confidence from it. Customers subject to rules on emotion recognition, such as the EU AI Act, should review that setting before using it.

Telling people it is an AI

Before a participant starts a conversation, the session’s notice tells them it is an AI tutoring session. The avatars are AI-generated video of stock presenters licensed through HeyGen, the voices are synthetic, and personas are characters an instructor writes, not real people. Our Terms do not allow anyone to use TutorTalks to pass an avatar off as a real person.

AI output is a draft, not a decision

AI output can be wrong, incomplete or biased. Insights and anything else the AI writes about a participant are drafts for the instructor to review. We do not use AI to make decisions about anyone. Our Terms forbid customers to use AI output in deciding whether a person is admitted or enrolled, selected for a program or opportunity, hired, promoted, paid, disciplined, dismissed, licensed or credentialed, or given or refused access to a service, and forbid using it as the sole or a principal basis for a grade of record or any other decision with legal or similarly significant effects on a person.

6. Who receives it

We disclose personal information only as described here. We do not sell it, and we do not share it for cross-context behavioural advertising. Our Subprocessors page names each provider, what it receives and where it processes data, and notes how long a provider keeps data where the provider publishes it.

RecipientWhat they receiveWhy
The customer that runs a sessionParticipant records for its sessions, by role: owners and administrators see every session in the organization, members (the role instructors usually have) see only the sessions they created, and billing administrators see figures but not participant names or conversations. Older sessions, created before TutorTalks began recording which member created each session, can be opened by everyone in the organization.To provide the service the customer pays for.
The participantTheir own copy of insights, if the instructor releases it, at a private link.So participants can see their feedback.
Cloud hosting: Google Cloud and Firebase, and HostingerEverything we store or process. Google runs our database, file storage, sign-in, background functions and analytics; Hostinger hosts our application server.To host and run TutorTalks.
AI providers: OpenAI, Google (Gemini), HeyGen, and ElevenLabs where selectedWhat section 5 describes.To run conversations and AI features.
Payments: StripeBilling details, our account and user IDs, your organization’s name on invoices, and the payment details you enter on Stripe’s page; for purchases made through Stripe Checkout, also the versions of our Terms and this policy you accepted and the ID of your acceptance record (section 7). Stripe and its affiliates process the payment transactions, and Stripe also uses this information under its own privacy policy, for example to prevent fraud.To take payments.
Networks that deliver page software: jsDelivr, Google and HeyGenThe IP address and browser details of people whose browsers load software libraries from jsDelivr, the Firebase software from Google, or avatar preview images from HeyGen on instructor pages.To deliver the pages.
Email deliveryThe recipient’s email address and the message, for credit links and account messages we send. Firebase sends password-reset emails.To deliver email.
Websites that embed a sessionWhen a participant joins an embedded session, the embedding page is told the session’s title and when it starts, ends or fails. It is never told the participant’s name, email address or conversation. The embedding website’s own practices are its owner’s responsibility.To fit the session into the page.
Advisers, authorities and others where requiredWhat is needed: to our lawyers, accountants and auditors; to comply with law or valid legal process; or to protect someone’s safety or our rights. Where the law allows, we tell the affected customer before disclosing participant records.Legal and safety reasons.
A buyer or successorPersonal information that transfers in a merger, acquisition or sale of assets. It stays subject to this policy and to our agreements with customers.A business transfer.
Tool servers an organization connects itselfOnly what the tutor puts into a tool request. This feature was withdrawn from sale and hidden in the dashboard in September 2026. An organization that attached tool servers before then may still have them in use; it chose them itself, and they are not our subprocessors.On the organization’s instruction.

7. Proof that you accepted

Before your first payment under each version of our Terms of Service and this Privacy Policy (a plan, credits, an AI-credit pack, a plan change or an add-on), TutorTalks asks the person paying to read and accept both. If your earlier acceptance was for a one-off purchase, we ask again before your first subscription, plan change or add-on. When you accept, we keep a record on our servers. Records are only ever added to; what is written is never edited afterwards.

What a record contains

  • Your user ID and, where there is one, your organization’s ID.
  • Your email address, if you have verified it.
  • The version of each document you accepted, and a SHA-256 fingerprint of each document’s text.
  • The statements you ticked (that you agree to both documents, that you are 18 or older and, if you are buying for an organization, authorised to accept for it, and, for a subscription, that it renews automatically every month until you cancel), and whether you scrolled to the end of the documents.
  • The date and time, from our server’s clock.
  • Your IP address and your browser’s user-agent string.
  • Which purchase prompted the acceptance, and a reference for the request, used to match it to our logs.

Each payment made under an acceptance, including the first, adds a short entry to the record: which purchase it was, the purchase summary the dialog showed if it was shown for that payment, the date and time, your IP address and browser user-agent string, a request reference and Stripe’s reference for the payment. These entries are kept for as long as the record.

We also archive the exact text of every version of these documents, so a record can be matched to the words you accepted. For purchases made through Stripe Checkout (plans, credits and AI-credit packs), we add the record’s ID and the document versions to Stripe’s record of the purchase. An acceptance may also be noted in your organization’s audit log.

Why we keep it, and for how long

We keep these records to prove that our agreement was made and on what terms, to answer payment disputes, and to establish, exercise or defend legal claims. Some automatic-renewal laws also require us to keep proof of consent. We keep each record for the life of the account plus six years, the time limit for contract claims under New York law, which governs our Terms. Because we need the records for legal claims, we do not delete them when an account is closed or when you ask us to delete your information, and we use them for no other purpose.

8. Cookies and browser storage

Our server does not set cookies. Our pages store some information in your browser, and Google Analytics sets cookies on four pages.

WhatWhereWhyHow long
Google Analytics cookies (named _ga and _ga_ followed by an ID) and related browser storage/pricing, /instructor, /studio and /redeemMeasuring how these pages are usedUp to two years after your last visit, unless you clear them
Firebase Authentication storage (a browser database)/pricing, /instructor, /studio and /redeemKeeping you signed inUntil you sign out or clear your browser data
Preferences (local storage)Instructor, studio and session pagesRemembering settings such as hands-free mode, your audio output device and the transcript layout, and the dashboard sign-in on passcode-protected installationsUntil you clear your browser data
Session storageSession and instructor pagesHolding a participant’s join ticket, including the pronoun answer, and the state of a purchase in progressUntil the tab is closed

Your choices

TutorTalks uses no advertising cookies, tracking pixels or session-recording tools, and Google Analytics is not loaded on the pages participants use, including embedded sessions. To stop Google Analytics you can block cookies and site data for tutortalks.io in your browser, use a content blocker, or install Google’s opt-out browser add-on. Blocking Firebase storage signs you out whenever you close the page.

Because we do not sell or share personal information or use it for targeted advertising, we already treat everyone as opted out of those, and a Global Privacy Control signal changes nothing further. These signals, and Do Not Track, do not currently stop Google Analytics, so use the options above if you want it off.

9. How long we keep it

We keep personal information only as long as we need it for the purposes in section 4. The periods below are what TutorTalks does today. Participant records follow the customer’s settings, which an owner or administrator of the organization can change in the dashboard.

InformationHow long we keep it
Participant audio recordings, and the transcript files archived with themUp to the end of the organization’s audio period: 90 days by default, and settable from 1 to 365 days. A daily process deletes them when the period ends. Recording can be switched off for a session or for the whole organization, and an instructor can delete a recording sooner.
Participant transcripts, visit details and AI-written insightsUntil the end of the organization’s transcript period: 365 days by default, and settable from 1 to 1,825 days (five years). The same daily process deletes them.
Participant records a customer deletesA recording the customer deletes is removed at once, and so are a participant’s records that we delete at the customer’s request. Deleting a session removes its transcripts, visit details and insights at once, but not yet the audio recordings and archived transcript files stored for its visits. To have those removed as well, ask us at studio@arclightxr.com, before or after deleting the session, and we will remove them within 30 days.
ExportsGenerated when requested and not stored by us; a temporary file is deleted as soon as it has been sent. A copy someone downloads, including an insights PDF a participant downloads, is outside our control.
Pronoun answers, in-session notes and live audioNot stored. They are held only in the browser tab or in our server’s memory while the session runs. The one exception is the coach’s one-word reading in our server log, kept for 30 days (section 3).
Approved-email listsUntil the instructor edits the list or deletes the session.
Sessions, knowledge bases, uploads and backdropsUntil the customer deletes them or the account is deleted.
Account, profile and organization information, and our account notesWhile the account exists. After we confirm a request to delete an account, we delete this information within 30 days, apart from the records below that we keep for longer.
Usage and cost recordsFor the life of the account. They hold counts and costs, with a one-way code in place of each session visit, not names or conversations. Figures we need for tax and accounting are kept with our billing records.
Security and audit recordsAt least two years; they are not deleted automatically at present. Because they show who accessed participant records, they outlast the records themselves, and an entry for exporting or deleting a participant’s records includes that participant’s email address.
Acceptance records, their payment entries and archived document textThe life of the account plus six years (section 7).
Billing records, including credit-link recordsAt least as long as tax and accounting rules require (generally seven years); they are not deleted automatically at present. Stripe keeps its own records under its own policies.
Server logs, including sign-in attempts and error reports30 days.
Google Analytics dataKept by Google for the retention period set on our Analytics property, which is no more than 14 months.
Support emailsAs long as needed to deal with the request and keep a record of what was agreed.

Deleting information removes it from TutorTalks’ live systems straight away. We do not keep scheduled backup copies of TutorTalks data, and our cloud provider completes erasure from its own systems on its own schedule. The one exception is the earlier copies of the few accounts and sessions created before 2 September 2026, kept from when we moved TutorTalks to its current database. Those copies are outside the periods above; we apply deletion requests to them as well, and we will delete them entirely.

10. Security and incidents

  • Connections to TutorTalks are encrypted (TLS, with HTTP Strict Transport Security), and Google encrypts the information we store with it at rest.
  • Our database and file storage refuse all direct access from browsers. Only our server can read them, and it serves a recording or transcript only after checking that the person asking is allowed to see it. There are no public links to files.
  • Each organization’s information is kept separate, and that separation is checked on the server for every request.
  • Roles limit who sees what: a member (the role instructors usually have) sees only the sessions they created, and a billing administrator cannot open a transcript. The exception is older sessions created before TutorTalks began recording which member created each session, which everyone in the organization can open.
  • Viewing, playing, downloading, exporting and deleting participant records, and changes to roles, retention and billing, are recorded in an audit log.
  • Our application logs are scrubbed of email addresses, sign-in tokens and keys, and are designed to exclude transcripts, audio and participant names. Our web server’s access log is the exception described in section 3: it records the address of each page requested, including any code or search text in it.
  • Access by Arclight personnel is limited to authorised people who need it to run, support or secure the service. Our internal operator tools cannot open transcripts or recordings, and we do not read a participant’s conversation unless the customer asks us to, the law requires it, or we need to investigate abuse or a security incident.

No system is perfectly secure. If you find a vulnerability, please email studio@arclightxr.com with “security” in the subject rather than disclosing it publicly, and we will work with you.

If there is a breach

If we discover a security breach affecting personal information, we will contain and investigate it and notify affected customers without undue delay. For participant records we hold for a school, college or university, we will notify the institution no later than seven calendar days after we discover the breach. We will also notify individuals and regulators where the law requires, telling them what happened, what information was involved, what we are doing about it and what they can do. Participants are usually notified by the customer that holds their records, and we will help.

11. Your rights and choices

Account holders, organization members and billing contacts

You can ask us to:

  • tell you what personal information we hold about you and how we use it;
  • give you a copy of it, in a portable format where practical;
  • correct it;
  • delete it, or close your account.

These requests are handled by email rather than in the dashboard. Write to studio@arclightxr.com from the email address on the account. We will confirm we have your request and respond within 30 days. Some things you can do yourself: cancel a subscription from the Account page; delete a session, which deletes its transcripts and visit details (section 9 explains what happens to its recordings); delete an individual recording; download a session’s records, where your organization allows downloads; and, as an owner or administrator, set your organization’s retention periods and switch recording off. To export or delete one participant’s records, an owner or administrator can ask us and we will do it for them within 30 days.

Deleting an account also deletes the customer’s participant records, so we will confirm with the account owner and offer an export first. If you are a member of someone else’s organization, the sessions you created there belong to that organization, and we may need its owner’s agreement to delete them. We keep the records listed in section 9 that we must keep, such as acceptance and billing records. We will not treat you differently for exercising your rights.

Participants

Participant records belong to the customer that ran the session, so requests to see, correct or delete them go to the organization or instructor that invited you. If we receive a request from a participant, we pass it to the customer, tell the participant we have done so, and help the customer respond. We do not change or delete a customer’s records without its instruction unless the law requires us to.

12. US state privacy rights

Residents of California and of other states with comprehensive privacy laws have rights over personal information about them. We give the disclosures below whether or not a particular law applies to us. For participant records we act as a service provider or processor, so those requests go to the customer, as section 11 explains.

Your rights

  • Know and access: the categories and specific pieces of personal information we hold about you, where it came from, why we use it and whom we disclose it to.
  • Delete and correct it, and receive it in a portable format.
  • Opt out of the sale or sharing of personal information, targeted advertising, and profiling in furtherance of decisions with legal or similarly significant effects. We do none of these.
  • Limit the use of sensitive personal information. The only sensitive personal information we collect for our own purposes is account sign-in credentials, used solely to sign you in.
  • Not be discriminated or retaliated against for using these rights, whether you are a customer or a participant.
  • Appeal if we decline your request.

How to ask. Email studio@arclightxr.com with “Privacy request” in the subject; you do not need an account. We verify a request by confirming that you control the email address our records hold for you, and we may ask for more information if that is not enough. An authorised agent can ask for you with your signed permission, and we may ask you to confirm your identity with us directly. We confirm receipt within 10 business days and respond within 30 days; if the law allows more time for a complex request, we will tell you why and when to expect an answer.

Appeals. If we decline all or part of your request, reply with “Appeal” in the subject. We respond to appeals within 45 days. If we deny your appeal, you can contact the attorney general of your state.

Other disclosures. We do not sell or share personal information, including that of anyone under 16. We do not disclose personal information to third parties for their own direct marketing. We do not collect, use or sell personal data for the purpose of training large language models. We do not use or disclose sensitive personal information to infer characteristics about anyone.

California notice at collection

In the past 12 months we have collected the categories of personal information below. We have not sold or shared any of them, and we have disclosed them for business purposes only to the recipients listed. Participant information appears here for completeness; we process it for our customers.

CategoryExamples and sourcesPurposes and recipientsHow long we keep it
IdentifiersName, email address, user and account IDs, IP address. From you, Firebase Authentication, your browser, and customers (for participants).Providing the service, security, billing, acceptance records and support. Hosting providers, Stripe, email delivery; AI providers receive a participant’s typed name, but not the email addresses we hold.For the life of the account; logs 30 days; acceptance records the life of the account plus six years.
Customer records (Cal. Civ. Code § 1798.80(e))Name, email, billing address, tax ID, payment details. From you, through Stripe.Billing. Stripe and hosting providers.At least as long as tax and accounting rules require, generally seven years.
Commercial informationPlans, purchases, credits and usage. From you, our systems and Stripe.Billing, metering and improving the service. Stripe and hosting providers.For the life of the account; billing figures at least as long as tax and accounting rules require.
Internet or other network activityServer logs, analytics on four pages, error reports, audit records. From your browser and our systems.Security and understanding page use. Hosting providers, Google Analytics, the networks that deliver page software, and HeyGen (a participant’s IP address and browser details).Logs 30 days; analytics up to 14 months; audit records at least two years.
Geolocation dataApproximate location (such as city) that Google Analytics derives from an IP address. No precise location.Understanding page use. Google.Up to 14 months.
Audio and electronic informationParticipants’ voice and recordings, for customers; an instructor’s voice in previews, which is not stored. From the microphone.Running the conversation and keeping recordings for the customer. AI providers and hosting providers.The organization’s audio period: 90 days by default, up to 365.
Professional or employment-related informationOrganization, role and membership. From you and your organization.Providing the service and access control. Hosting providers.For the life of the account.
Education informationTranscripts, visit details and insights held for schools, colleges and universities. From the conversation.Providing the service to the institution. AI providers and hosting providers.The organization’s transcript period: 365 days by default, up to five years.
InferencesAI-written insights about a participant’s session, for customers with the Student insights add-on. Produced by the language-model provider.Feedback for the instructor and, if released, the participant. AI and hosting providers.The organization’s transcript period.
Sensitive personal informationAccount sign-in credentials (email address and password), held by Firebase Authentication.Signing you in, and nothing else. Google (Firebase).Until the account is deleted.

13. EEA, UK and Swiss residents

If you are in the European Economic Area, the United Kingdom or Switzerland, data protection law gives you these additional details and rights.

  • Controller. Arclight Innovation LLC is the controller of the information described in section 2 as ours, and can be reached at studio@arclightxr.com. We have not appointed a data protection officer. If we appoint a representative in the EU or the UK, we will name them here. For participant records and personal information in Customer Content, the customer is the controller and we are its processor.
  • Legal bases. Section 4 sets out the basis for each purpose. Where we rely on legitimate interests, we have weighed them against your rights, and you can ask us for details. We do not rely on consent for any processing described here; if we ever ask for it, you will be able to withdraw it at any time.
  • What is required. You must give us account and billing details to buy and use TutorTalks. A participant must give an email address to join a session; everything else they provide is optional or part of the conversation itself.
  • Your rights. You can ask for access to your information, its rectification or erasure, restriction of processing, and portability, and you can object to processing based on legitimate interests, including analytics. We do not make decisions about you based solely on automated processing that have legal or similarly significant effects. We respond within one month, or tell you within that month if we need up to two more because a request is complex.
  • Complaints. You can complain to us at studio@arclightxr.com; we acknowledge complaints within 30 days and tell you the outcome. You also have the right to complain to a supervisory authority, such as the data protection authority where you live or work, the Information Commissioner’s Office in the UK, or the Federal Data Protection and Information Commissioner in Switzerland.
  • Transfers. Section 15 explains how information reaches the United States and the safeguards we use.

14. Children and age limits

TutorTalks is not directed to anyone under 18. Account holders must be 18 or older. Participants must be 18 or older unless we have agreed otherwise in writing with the customer, because several of the AI providers TutorTalks depends on require their users to be adults.

We never knowingly collect personal information from a child under 13. If we learn that we have, we will delete it and tell the customer concerned. If we learn that someone under 18 has taken part without such an agreement, we will tell the customer, who must stop that use under our Terms, and we may suspend the session; we delete that participant’s records on the customer’s instruction or where the law requires. If you believe either has happened, please email studio@arclightxr.com. A K-12 school must contact us before using TutorTalks, and may use it only under a written agreement with us.

15. Where data is processed

Our own infrastructure is in the United States: our application server is in Boston, Massachusetts; our database runs in Google Cloud’s United States multi-region; and our background functions run in Google Cloud’s us-central1 region. Our providers may process information in other countries, and the Subprocessors page says where each one does.

If you use TutorTalks from outside the United States, your information is processed in the United States, where data protection law may differ from yours. Where the law of the EEA, the UK or Switzerland requires a transfer safeguard, we rely, for information our providers process for us, on the European Commission’s standard contractual clauses and the UK’s International Data Transfer Addendum included in those providers’ data processing terms. For participant records and Customer Content that an organization in the EEA, the UK or Switzerland puts into TutorTalks, those clauses apply once that organization and Arclight have signed a data processing agreement that includes them, and such an organization should sign one with us before using TutorTalks with participants. We are not certified under the EU-US Data Privacy Framework. Any institution that needs a data processing agreement can ask us for one.

16. Changes to this policy

When we change this policy, we publish the new version here with a new version date, and we keep every earlier version, which we will send you on request. For a material change, we will email account owners at least 30 days before it takes effect, unless the law or an urgent security need requires a change sooner. The dialog shown before payments will ask you to accept the new version before your next purchase.

Changes apply going forward. We will not use information we have already collected in a materially different way without your consent, and because we act only on our customers’ instructions for participant records and Customer Content, a change in how we use them applies to a customer only once it has agreed to it. Changes to the providers we use are announced on the Subprocessors page.

17. Contact

TutorTalks is operated by Arclight Innovation LLC. Write to studio@arclightxr.com about this policy, your information, or a security concern. Putting “Privacy request”, “Appeal” or “Security” in the subject helps us route your message.

If you are a participant: requests about your session records go first to the organization or instructor that invited you; if the session is run by a school, your instructor or the registrar’s office is the right contact. If you write to us, we will pass your request on and help them respond.