Skip to content

Privacy Policy

Document version: 2026-11-10.1
Effective date: 2026-11-10

How we use your data

This is a short summary of the Privacy Policy below. Each answer links to the full section. If this summary and the policy differ, the policy applies.

What do you collect when I sign up and apply?
Your account details (email, name, phone, country and language), your date of birth if you create a freelancer account, and what you give us to apply: your résumé, skills, languages, work history, links and a profile photo. Some jobs also ask for an ID and selfie check, your location, or a recorded interview. We also collect technical data: your country from your IP address, VPN or proxy signals, device and security events, and error reports, which can include limited, masked recordings of your session. We keep the messages you send our support team, including on WhatsApp. Website analytics run only if your cookie choice allows them. [§1]

Who sees my profile and résumé?

  • Employers and clients: when you apply, the employer or client for that job can see what the application needs, such as your profile, résumé, screening answers and interview information. Depending on your settings, some profile details may be public or searchable. [§3, §10]
  • Service providers: companies that run parts of the service for us, such as storage, AI and speech processing, identity checks, support and email, get only the data their task needs. The policy lists each one, what it receives and what it may keep. [§3]
  • When the law requires it: for safety, for a legal requirement, or if the company is sold or merges. [§3]

Does an AI read my résumé or my interview?
Yes. AI models read your résumé to fill in your profile, and we keep a searchable version of it to match you to work. In an interview, a speech service turns your answers into text and reads the questions aloud, and AI models process your transcript and answers. Automated checks also look at the name you sign up with and at messages on OpenTrain, to catch scams. These checks never ban anyone by themselves. AI does not make hiring decisions on its own: employers and OpenTrain staff decide. You can ask for a person to review an important automated result. [§2, safety checks, §3, §5]

Do you use my data to train AI?
OpenTrain does not use your personal data to train its own general-purpose or foundation AI models. To read résumés and run interviews, we send your résumé, profile, interview audio and transcript to AI model and speech providers. The policy lists each provider and whether it may use that data to improve its models. Deepgram’s opt-out has been on since October 20, 2026: current requests carry the opt-out and are not used for Deepgram’s model-improvement program. Before that date, interview audio was included in that program and Deepgram could use it to improve speech models in exchange for a 50% discount. That earlier use may be treated as a sale or sharing under some privacy laws. Use the Do Not Sell or Share My Personal Information link to opt out of any applicable historical use. [§2, §3]

Do you sell my data?
We do not sell your personal data for money as a standalone product. Two things may still count as a “sale” or “sharing” under some privacy laws, such as California’s:

  • Deepgram’s historical use of your interview audio to improve its speech models before October 20, 2026, described above; current requests carry the opt-out.
  • LinkedIn server-side conversion measurement for LinkedIn ad clicks. After you allow analytics/attribution cookies, when you sign up or click through to a partner from a LinkedIn ad, we may send LinkedIn the LinkedIn first-party ad click identifier, the conversion time and an opaque event ID. We do not send your name, email address, IP address or job data. This may count as a “sale” or “sharing” under some privacy laws.

To opt out, see Do Not Sell or Share My Personal Information: use the privacy controls in the cookie banner, or email support@opentrain.ai with the subject “Privacy request”. We also honor the Global Privacy Control browser signal where the law requires it. [§3, opt-out, §11]

Why do you ask for ID or a selfie?
Some jobs need us to confirm that you are the person applying, to keep accounts secure and to prevent fraud. An identity provider (Didit or Stripe Identity) compares your selfie with your ID document. This is a biometric check. OpenTrain keeps only a redacted, non-biometric record of the result, as evidence of the check. We do not keep your ID document, your selfie or a face template. [§1, §3, §4]

Is the interview recorded, and how long do you keep it?
Yes, when a job includes a recorded interview. We record your camera and microphone and keep the transcript and your answers. During the interview we run integrity checks: tab switches, virtual cameras, read-aloud and speech checks, and a comparison of your voice across answers to check that one person completes the interview. We do not keep a voiceprint. We delete recordings 12 months after the interview, and detailed integrity-check results 180 days after it. The overall outcome, such as whether a session was flagged, stays on your account record. [§1, §4]

Where is my data stored?
With our database and storage providers: Supabase for account, profile and interview data, and Amazon S3 for archived interview recordings. These and our other providers may process your data in a country other than yours. We use the contractual and technical safeguards the law requires. [§3, §6]

How long do you keep my data?

  • Account and application data: while your account or application is active, then only as long as needed for legal, safety, fraud and dispute reasons.
  • Interview recordings: 12 months after the interview. Detailed integrity-check results: 180 days after the interview.
  • Unfinished application answers: deleted after 30 days.
  • Results of automated name and message checks: usually up to 90 days, longer while a related review is open. After that we keep only overall counts.
  • ID check: we keep the redacted result as evidence of the check.
  • Logs, diagnostics, analytics and support records: as long as needed for security, running the service, reporting and legal duties, then deleted or de-identified.
  • Your consent choices: kept as a record of what you chose and which policy version applied.

[§4, safety checks]

How do I ask to delete my data?
Email support@opentrain.ai with the subject “Privacy request” and include your account ID. Do not send ID documents by email. We may need to confirm it is you. You can also ask to see, correct, export or limit your data, or object to how we use it. If you delete your account, we delete your interview recordings and detailed integrity-check results within 30 days. We may keep some data where the law allows or requires it. [§4, §5]

How do I report a problem?
Email support@opentrain.ai with the subject “Privacy request”. If you think a message was held or your account was limited by mistake, email the same address and we will review it. You can also complain to the data-protection authority where you live, for example the ANPD in Brazil or the Agencia de Acceso a la Información Pública in Argentina. [safety checks, §5]

Full Privacy Policy

OpenTrain AI (“OpenTrain”, “we”, “us”) provides recruiting, annotation and interview services. OpenTrain AI is a trade name of WMD Global, LLC, the controller for personal information processed under this Privacy Policy. Mailing address: OpenTrain AI
5725 S Valley View Blvd, Ste 5, PMB 20762
Las Vegas, NV 89118-3122, United States. This Privacy Policy explains what we collect, why we use it, where it goes, how long we keep it, and the choices available to you.

1. Data we collect

  • Account and contact data: email, name, user type, phone number, country, language, sign-in and support information.
  • Profile and application data: resume files and parsed resume text, skills, languages, employment and application information, links, profile photo and employer screening information.
  • Identity and safety data: identity-document and selfie checks performed by an identity provider; location, GPS when a job requires it, IP-derived country and VPN or proxy signals; browser, device and security events.
  • Interview data: camera and microphone audio/video, transcript, answers and interview metadata. We run integrity checks such as tab-switch detection, virtual-camera checks, read-aloud and speech checks, and comparison of the voice across answers to check that one person completes the interview.
  • Biometric information: a provider may compare a selfie to an identity document, and the interview service may compare voice characteristics across answers. OpenTrain keeps a redacted, non-biometric verification result snapshot; it does not keep the provider’s identity documents, selfie template or voiceprint.
  • Customer support communications: support tickets, emails, call notes, and chat transcripts, including messages you send our support team on WhatsApp.
  • Usage and diagnostics: error data, internal user ID and role, limited masked session replay, page and performance measurements, support events, and website analytics or attribution events when the applicable consent choice permits them. On the website, PostHog may capture a masked replay of a page that is slow or fails; those replays are kept for 30 days and are disabled when analytics cookies are rejected or Global Privacy Control is sent.
  • Surveys. We may ask you one short, optional question inside the OpenTrain app, not on our website—for example, how easy it was to post a job. PostHog runs the survey for us, and it appears only if you allowed analytics in the app’s privacy settings. Your answer is linked to a coded account ID, not to your name or email. We use it only to improve the product. PostHog’s project setting expires survey answers after 12 months; deletion may run in batches after expiry.
  • Cookies and similar technologies: the Cookie Policy below lists the current cookies, analytics and advertising tags.

Date of birth

If you create a freelancer account, we ask for your date of birth during onboarding, after you upload your resume. Until you pass the age check, we do not use your resume or the profile details filled in from it for job matching, search or job alerts, and we do not show them to employers. If you have not given us your date of birth, we ask for it before you start an identity verification.

  • Why we collect it: We use your date of birth only to check that you are 18 or older and to match it with the date of birth on your ID when a job or partner requires identity verification.
  • If you are under 18: Your account is locked and we do not keep the date you entered. We delete your resume and the profile details filled in from it right away. The same happens if an identity verification shows that you are under 18. Until the account is deleted, we keep only your name, email address and country, the result of the age check and when it happened. The account is deleted 30 days later unless an identity verification shows that you are 18 or older before then. If it does, you can upload your resume again.
  • Identity match: During an identity verification, the date of birth on your ID is compared with the one you gave us. We keep only the result (match or no match), not the date of birth on your ID. If the dates do not match, we review the verification. If we approve it, the date of birth on your ID replaces the one you gave us.
  • Who sees it: In the Services, only you can see your date of birth. Our support team sees only whether the age check and the ID match passed. We never use your date of birth for job matching, and we never show it to employers, partners, or on your profile.
  • How long we keep it: We keep your date of birth while your account is open and delete it when your account is deleted. Copies in our database backups are deleted when those backups expire.

Business data matching

When you use our check-eligibility form, we collect these categories of information:

  • company details you provide, including the company name, website, operating status, country, size, and the kinds of records the company has; and
  • your name, role, and work email.

We use this information to check which AI data buyers fit the company, send the details to the buyers that fit so they can contact you about the company, and email you about the matching and what happens next.

If you are outside the United States, we use the details you send to check which AI data buyers fit, send them to the buyers you tell us to, and email you the result because you asked us to when you sent the form. To object or stop, use the unsubscribe link in any of our emails, or write to support@opentrain.ai with the subject line “Privacy request.”

Do we sell your personal information? No. We send your details to an AI data buyer only because you tell us to when you submit the form. A buyer may pay us a referral fee for the introduction. You pay nothing. To stop us from sending your details to any buyer, use Do Not Sell or Share My Personal Information. We check this, and your browser’s Global Privacy Control signal, right before we fill each buyer’s form.

We do not collect sensitive personal information through this form.

We keep completed form details and related records for as long as reasonably necessary to complete the matching and follow-up, enforce agreements, resolve disputes, prevent fraud and abuse, and meet legal, accounting, and security obligations. The retention period therefore depends on the purpose and the applicable obligation, as described in Section 4 of this Privacy Policy. If you start but do not finish the form, we delete the saved answers 30 days after your last step, except that we keep your email address on our do-not-email list when you unsubscribe, our email bounces, or you mark it as spam.

We honor Global Privacy Control. If your browser sends it, we don’t send your details to any buyer unless you tell us yes by email for that buyer. We check it again right before each buyer form is filled.

2. How we use data

We use data to create and secure accounts, match people to work, parse resumes, run interviews and integrity checks, verify identity and location, provide support, prevent abuse, measure performance, communicate with you, operate application and buyer-introduction forms, and comply with law and improve the service. Some forms save answers at each step; unfinished answers may trigger up to three reminders and are deleted after 30 days. If you choose a buyer introduction, we send the selected buyer the company details, your full name and your work email, and OpenTrain may receive a referral reward from that buyer.

AI training answer. OpenTrain does not use freelancer personal data to train its own general-purpose or foundation models. OpenTrain sends resume, profile, interview audio and transcript data to model or speech providers to provide those features. Each provider row below states the applicable training and retention position. OpenTrain does not sell personal data as a standalone product. Deepgram’s opt-out has been on since October 20, 2026: current requests carry the opt-out and are not used for Deepgram’s model-improvement program. Before that date, interview audio was included in that program and Deepgram could use it to improve speech models in exchange for a 50% discount. That earlier use may be treated as a sale or sharing under some privacy laws. Use the Do Not Sell or Share My Personal Information link to opt out of any applicable historical use.

Safety, security, and integrity checks

We use your first and last name to run an automated check using an AI service provider. The check may flag a name that may not be a legal name. A flag by itself does not block or change your account. The service provider that runs this check keeps no copy of your name. If the check flags your name, the flag goes to our trust and safety review, which uses a separate AI service provider; that provider receives your name and related review details under its standard terms, which allow it to keep that data for a limited period. We keep the check record for 90 days and then retain only aggregate counts. This applies when you sign up and when you edit your profile name, in every country. If a check flags your name, you can ask for a human review or appeal the flag through our support process.

Work Email safety checks. When an OpenTrain work email sends or receives messages about an OpenTrain job, proposal or contract, we may automatically check new messages for scams and for attempts to move contact, work or payment off OpenTrain. This can include messages from a client or colleague who does not have an OpenTrain account. We do not check other mail. An AI service provider helps with the check and keeps no copy of the message. The check keeps no copy of the message either; we keep its result for up to 90 days, or longer only while a review of it is still open. A check never bans or removes anyone by itself. A person reviews what it finds and may read the flagged part of the message when needed to decide.

Chat message safety checks. When this feature is enabled, OpenTrain automatically checks messages that people send to each other in OpenTrain, before or after a hire, including direct messages, team channels, thread replies, edits, mobile messages and API messages, for scams, abuse, requests to pay or be paid outside OpenTrain, attempts to collect contact details, and attempts to move the conversation or the hiring off OpenTrain.

We check messages from both sides. The check reads the message as written, which can include contact details that a user typed, and the least context it needs to understand it: recent messages in the same conversation and, when available, the job or proposal title and a short description, hiring headcount, the employer account’s trust level, recent conversation-count context, and the names of attached files.

For a message after a hire, the check is told that normal work tools, such as the client’s labeling tool, meeting links and file shares, are part of the work; those tools alone are not treated as an attempt to move work off OpenTrain.

When an employer attaches files, the check may also read the text on up to three readable files, which can include personal details in the file, with a limit of 8,000 characters per file and 16,000 characters total. Candidate attachments are never read. Scanned or image-only files are not read. The extracted file text is sent only to the pinned safety-check models through providers configured for zero data retention and denied data collection, with no fallback provider. It is treated as untrusted evidence, not instructions. The check’s records do not keep the file text, file name, path or link; they keep only fixed codes showing whether each file was read or could not be read.

AI service providers help with the check and keep no copy of the messages or extracted file text. The check does not keep a copy of the message either; we keep its result for up to 90 days, or longer while a related hold, review or account limit is still open and for a limited time after it ends, and then keep only overall counts. Some check records made before October 27, 2026 that are tied to an open review may still hold a copy of the message until that review ends or January 26, 2027, whichever comes first.

If a check finds a likely problem, we may hold or block the message, or limit what the employer’s account can do, and our trust and safety review looks at what the check found. A held message is dropped and cannot be delivered later. If a file is held as review evidence, only an authorized OpenTrain AI Admin reviewer can open it through an audited route; it is deleted within 30 days after the review decision or within 60 days after it was held, whichever comes first. A candidate’s message alone never limits anyone’s account, and a check never bans anyone by itself. If you think a message was held or your account was limited by mistake, email support@opentrain.ai and we will review it.

Chat message scam checks. We may also automatically check other chat messages on OpenTrain for scams: messages between employers and freelancers about a job after a hire, direct messages that are not covered by the pre-hire checks above, including messages from candidates and freelancers, messages in group channels, replies in a message thread, and edited messages. The check looks for messages that ask someone to pay to get or keep work, to buy equipment, software or training from the sender, to take part in an overpayment or refund scheme, to pay in cryptocurrency or gift cards, or to share bank details, identity documents, logins or verification codes, and for messages that pretend to be OpenTrain or another company. Before the check, we remove email addresses, phone numbers and internal identifiers from the message. The check receives the rest of the message as written, including the address of any link in it, which can still include a name or other details someone typed, and only two other facts: whether the sender is an employer or a freelancer, and whether the conversation is about a job after a hire. It does not receive profile details, job details, earlier messages or attachments. An AI service provider runs the check through providers configured for zero data retention and denied data collection, and keeps no copy of the message. The check runs after a message is delivered. It never blocks, holds, delays or changes a message, and never limits an account. If it finds a likely scam, our trust and safety reviewers may open that message in OpenTrain, through their scoped review access and with no stored copy, to decide whether anything needs to happen; a check result alone is never the reason for an action on an account. The check’s records keep no message text. We keep each result for up to 90 days and then keep only daily counts; a result sent for review is kept while the review is open and for 90 days after it ends. Messages from OpenTrain’s platform operators and messages OpenTrain writes itself are not checked.

Job post safety checks. When an employer submits a job post for publication on OpenTrain, or edits a published one through our API, we may automatically check it for scams, requests to pay or be paid outside OpenTrain, contact details in the post (including requests to contact the employer outside OpenTrain), and empty or junk posts. Some checks are simple rules that run inside OpenTrain. For the rest, we remove email addresses, phone numbers and internal identifiers from the post, and an AI service provider checks the post’s title, description and other job details as written, which can still include a name or other details the employer typed. It does not receive the employer’s email address, the company name or any account identifier. The request goes to one AI model provider only, with no fallback to another provider, under settings for zero data retention and denied data collection, and no provider that handles the request keeps a copy of the post. The check never blocks, hides, delays or changes a job post, and never limits an account; our existing job post review still decides whether a post is published. If the check finds a likely problem, our trust and safety review, which uses an AI review agent, may look at the post as written, including any contact details in it, to decide whether anything needs to happen, and the standard terms of the AI provider behind that review agent allow it to keep the review input for up to 30 days, subject to exceptions in those terms; a check result alone is never the reason for an action on an account. The check’s records keep no job post text. We keep each result for up to 90 days and then keep only daily counts; a result sent for review is kept while the review is open and for 90 days after it ends. Partner jobs that OpenTrain shows from other sites are not checked.

Profile headline and bio checks. When you complete your freelancer or agency profile, and later when your profile headline or bio changes (for example when you edit it, or when a résumé you upload fills it in), we may automatically check them for contact details, outside links, gibberish or profanity. Contact details and links are looked for by simple rules that run inside OpenTrain. For gibberish and profanity, an AI service provider checks your headline and the first 4,000 characters of your bio. Before that, we replace the email addresses, phone numbers, messaging contacts, links and postal addresses that our rules recognize. We do not send your account details, such as your name, email address or country, as separate information, but the provider receives what you wrote in your headline and bio, which can include such details if you wrote them there. The request goes to one AI model provider only, with no fallback to another provider, under settings for zero data retention and denied data collection, and no provider that handles the request keeps a copy of the text. If the check finds a likely problem, our trust and safety review, which uses an AI review agent, may look at your current headline and full bio as written, including any contact details in them, to decide whether anything needs to happen, and the standard terms of the AI provider behind that review agent allow it to keep the review input for up to 30 days, subject to exceptions in those terms. The check and the review never hide, change or block your profile, and a check result alone is never the reason for an action on an account. The check’s records keep none of your headline or bio. We keep each result for up to 90 days and then keep only daily counts; a result sent for review is kept while the review is open and for 90 days after it ends.

Application answer checks. When you complete an AI interview for a job on OpenTrain, we may automatically check your screening answers for answers that are empty or placeholders, the same text pasted into several questions, and junk such as keyboard mashing. These checks never judge whether an answer is good, correct, fluent or relevant, and short answers, any language and spelling or grammar mistakes are never a reason to flag. Some checks are simple rules that run inside OpenTrain. For the rest, we remove email addresses, phone numbers, links and internal identifiers from your written answers, and an AI service provider checks the questions and those answers as written, which can still include a name or other details you typed. It does not receive your name, email address, country or resume, the job post, any account identifier, or your answers to questions that ask for a number, a link or a choice from a list. An answer you wrote in your own words is still sent if it contains numbers. The request goes to one AI model provider only, with no fallback to another provider, under settings for zero data retention and denied data collection, and no provider that handles the request keeps a copy of your answers. The check never blocks, hides, delays or changes your application, your interview, its score or what the employer sees, and never limits your account. If the check finds a likely problem, our trust and safety team may look at the application to decide whether anything needs to happen; a check result alone is never the reason for an action on an account. The check’s records keep no answer text. We keep each result for up to 90 days and then keep only daily counts; a result sent for review is kept while the review is open and for 90 days after it ends.

Company name checks. When an employer or agency gives a company name at sign-up, or an employer changes it in Settings, we may automatically check whether it looks like a placeholder (such as “test” or “none”) or an obviously fake name. We do not check it if you choose to keep your company private. Some checks are simple rules that run inside OpenTrain. For the rest, an AI service provider checks the company name only, with any email address or phone number in it removed. It does not receive your name, email address, country or any account identifier. The request goes to one AI model provider only, with no fallback to another provider, under settings for zero data retention and denied data collection, and no provider that handles the request keeps a copy of the name. The check never blocks or delays a sign-up, never changes the name you saved, and never limits an account. If the check finds a likely problem, our trust and safety team may review it, and a check result alone is never the reason for an action on an account. The check’s records keep no company name. We keep each result for up to 90 days and then keep only daily counts; a result sent for review is kept while the review is open and for 90 days after it ends.

We may use aggregated or de-identified service metrics that do not identify a person to improve reliability and product quality. We do not make employment decisions solely by automated processing; employers and OpenTrain staff make decisions using the information available to them.

For people in the European Union and the United Kingdom, we process personal data under these legal bases, depending on the purpose and the data involved:

  • Contract: to create and secure an account, provide the Services you request, run applications and interviews, provide support, process payments, and perform a managed-services or marketplace agreement.
  • Legitimate interests: to prevent fraud and abuse, protect account and platform security, run safety and integrity checks, provide and improve the Services, maintain service records, respond to disputes, and operate our business. We balance these interests against your rights and expectations.
  • Consent: for optional cookies, analytics, attribution and advertising, and for any processing where applicable law requires consent. You may withdraw consent at any time; withdrawal does not affect processing that happened before withdrawal.
  • Legal obligation: to keep accounting, tax, compliance and consent records, respond to lawful requests, and meet other duties imposed by law.

Where we rely on legitimate interests, you may object on grounds relating to your situation. Where we process special-category data, we rely on an applicable condition under Article 9 GDPR and explain or request consent where required.

3. Sharing and subprocessors

We share only the data needed for the purpose with service providers acting on our instructions, with employers or clients when an application or job requires it, and when required for safety, law or a corporate transaction. When you choose a buyer introduction, we send the selected buyer the company details, your full name and your work email, and OpenTrain may receive a referral reward from that buyer. Deepgram may receive interview audio for speech processing. Its opt-out has been on since October 20, 2026, so current requests are opted out of model improvement; before that date, interview audio was included in Deepgram’s Model Improvement Program and could be used to improve speech models in exchange for a 50% discount. That earlier use may be a sale or sharing under some privacy laws. Email and communications providers include Twilio, which sends verification codes, and WhatsApp, a Meta company, when you choose to get a code on WhatsApp or message our support team on WhatsApp. For LinkedIn ad attribution, after you allow analytics/attribution cookies, our server may send LinkedIn a first-party ad click identifier, conversion time and opaque event ID when you sign up or click through to a partner; it sends no name, email address, IP address or job data. No LinkedIn advertising tag runs in the browser. Use the Do Not Sell or Share My Personal Information link to opt out of any applicable historical use. We honor rejection of optional technologies and Global Privacy Control where required by law.

ProviderPurpose and dataTraining and retention
Supabaseaccount, application, profile and interview storage; database and file bucketsOpenTrain retention schedule applies
Amazon S3 (including Glacier Instant Retrieval)archived interview recordings and other filesrecordings move to archive after 30 days; the retention period below applies
OpenRouterrouter for model calls carrying resume, profile, job, message, interview text or audio dataOpenRouter routing sends a no-data-collection instruction to supported providers. OpenRouter’s optional use of inputs and outputs is off by default; the account setting has not been independently verified. Default abuse-monitoring retention may be up to 30 days. Request metadata such as token counts and latency is retained. No zero-retention claim is made.
Anthropic (direct API and when routed through OpenRouter)resume parsing, interview prompts and related text; trust and safety review of flagged content as written, which can include contact detailsNo model training under commercial API terms. Default retention is up to 30 days; policy-review content may be kept up to 2 years, classification scores up to 7 years, feedback up to 5 years, Files API uploads until deleted, and data longer where law requires it. No zero-retention claim is made.
Deepgramspeech-to-text for interview audio and text-to-speech for interview questionsOpt-out on since October 20, 2026: current requests carry the opt-out and are not used for model improvement. Before that date, interview audio was included in Deepgram’s Model Improvement Program and could be used to improve speech models in exchange for a 50% discount; Deepgram may retain it under program terms.
Pineconeresume or profile embeddings used for search and matchingreceives embeddings and identifiers needed for retrieval
Didit; Stripe Identityidentity document and selfie checksproviders receive the identity material needed for verification; OpenTrain keeps only a redacted non-biometric result snapshot
Sentryerror diagnostics and limited masked session replayUS service; replay follows app consent controls
Vercel Analytics and Speed Insightswebsite and app page and performance measurementsfollows app and website consent controls
Intercom; Chatwootsupport messaging and support contextreceives account and support data when support is used
Email delivery providerverification and service emailsreceives email address and message metadata
proxycheck.ioIP, country and VPN/proxy risk signalreceives IP-derived request data for location and abuse prevention
PostHogwebsite analytics, attribution and masked replay of slow or broken pages and optional in-app surveyswebsite collection follows consent and GPC; masked replays are kept for 30 days and are disabled when analytics cookies are rejected or GPC is sent
Probowebsite consent bannerstores consent choice and cookie preferences
LinkedIn Conversions APIserver-side conversion measurement for LinkedIn ad clicks; first-party ad click identifier, conversion time and opaque event ID onlyonly after analytics/attribution consent and only for a signup or partner click-through; no name, email, IP address or job data

We require providers to protect data and use it only for the contracted purpose. We maintain this table when providers change.

4. Retention

  • Account and application data: while the account or application is active, then as needed for legal, safety, fraud and dispute purposes.
  • Verification result snapshot: retained as historical evidence. It contains a redacted, non-biometric result, not the identity document, selfie or voiceprint.
  • Interview recordings, including audio, video and answer clips: kept for 12 months after the interview, then deleted. Recordings move from hot storage to S3 Glacier Instant Retrieval after 30 days. The same 12-month period applies to retained recordings from 717 accounts deleted before this retention rule took effect.
  • Detailed integrity-check results, including tab-switch signals and camera and voice-check detail: kept for 180 days after the interview, then deleted. The outcome itself, such as whether a session was flagged and its strike count, remains with the account record. The same 180-day period applies to retained detailed results from 717 accounts deleted before this retention rule took effect.
  • Unfinished application and buyer-introduction form answers: deleted after 30 days. A form may save answers at each step and may send up to three reminders before deletion.
  • Masked PostHog replays of slow or broken website pages: kept for 30 days, and disabled when analytics cookies are rejected or Global Privacy Control is sent.
  • Logs, diagnostics, analytics and support records: for the period needed for security, service operation, reporting and legal obligations, then deleted or de-identified under the applicable system schedule.
  • Date of birth: while the account is open, then deleted when the account is deleted. Copies in database backups are deleted when those backups expire. We do not keep the date of birth of anyone under 18 (Date of birth).
  • Consent records: retained as evidence of the choice and policy version.

If you delete your account sooner, we delete interview recordings and detailed integrity-check results within 30 days. We delete or de-identify data sooner when required by law or when a valid request applies, subject to lawful exceptions.

5. Your rights

You may ask us to confirm, access, correct, update, export, restrict or delete personal data, object to processing, withdraw consent where processing is based on consent, and ask for human review of a significant automated result. We may need to verify your request. You may complain to your local data-protection authority.

  • Brazil (LGPD): you may exercise confirmation and access, correction, anonymization, blocking or deletion of unnecessary or excessive data, portability, information about sharing, review of automated decisions, and consent withdrawal rights. You may complain to the ANPD.
  • Mexico: you may exercise ARCO rights (access, rectification, cancellation and opposition), withdraw consent and object to uses not covered by the notice.
  • Argentina: you may request access, correction, update or suppression and complain to the Agencia de Acceso a la Información Pública.
  • California and other US states: you may request access, correction, deletion, portability and information about disclosure or sale/sharing. Deepgram’s historical model-improvement use before October 20, 2026 is described above and may be a sale or sharing; current requests are opted out. Use the Do Not Sell or Share My Personal Information link and Global Privacy Control where applicable.

Do Not Sell or Share My Personal Information

This section explains the opt-out link. Use the privacy controls in the cookie banner or email support@opentrain.ai with the subject line “Privacy request” to ask us to stop sale or sharing where the law gives you that right. We honor Global Privacy Control where required by law. The request covers LinkedIn server-side conversion measurement that uses a first-party ad click identifier after analytics/attribution consent, and any applicable historical Deepgram model-improvement use before October 20, 2026; current Deepgram requests already carry the opt-out.

Privacy contact: email support@opentrain.ai with the subject line “Privacy request.” Include your internal account ID; do not send identity documents by email. No DPO or local representative is named in this version.

6. Security and international transfers

We use access controls, encryption and monitoring appropriate to the risk. Providers may process data in countries other than yours. We use contractual and technical safeguards required by applicable law. No online service can guarantee absolute security.

7. Changes

We will post the effective date and describe material changes. The English and Spanish versions are intended to have the same meaning; if they differ, contact us before relying on a translated provision.

What changed on November 8, 2026: profile headline and bio checks now say that a flagged profile may be looked at by an AI review agent as written, including any contact details in it, and how long that review input may be kept.

What changed on November 7, 2026: optional in-app surveys in Section 1 and in the PostHog row of the provider table.

What changed on November 6, 2026: profile headline and bio checks for freelancers and agencies.

What changed on November 6, 2026: company name checks at employer sign-up and in Settings.

What changed on November 5, 2026: a new paragraph describes application answer checks on AI interview screening answers.

What changed on November 5, 2026: job post safety checks now say that a flagged post may be looked at by an AI review agent as written, including any contact details in it, and how long that review input may be kept; the paragraph no longer names individual AI providers; and the provider table lists trust and safety review of flagged content.

What changed on November 4, 2026: we added the Date of birth entry in Section 1 (why we ask for it, what happens if you are under 18, the ID match, who can see it and how long we keep it), its retention line in Section 4, date of birth in the summary at the top, and matching Terms §3.1 wording.

What changed on November 2, 2026: we corrected the Deepgram model-improvement opt-out status and disclosed the earlier program use; added the EU/UK legal-bases section; added the Business data matching Notice at Collection and its direct opt-out link; automated name checks at sign-up and profile edits; Work Email safety checks; pre-hire message safety checks, including bounded employer attachment-text screening; chat message scam checks after a hire and on other messages the pre-hire checks do not cover; job post safety checks; support messages you send us on WhatsApp, a Meta company; and matching Terms §6.4 wording.

8. Managed services and customer content

OpenTrain provides a platform where freelancers create profiles, find opportunities, apply for jobs, communicate with employers, perform work and get paid; employers and clients post jobs, evaluate candidates, hire freelancers, manage work and pay freelancers; data-labeling vendors and agencies submit proposals and manage vendor relationships; and managed-service customers engage OpenTrain for recruiting, onboarding, project operations and quality assurance.

Some application and buyer-introduction forms save answers at each step. If a form remains unfinished, we may send up to three reminders, and we delete the unfinished answers after 30 days. If you choose a buyer introduction, we send the selected buyer the company details, your full name and your work email. OpenTrain may receive a referral reward from that buyer.

OpenTrain generally acts as a controller for account registration, platform use, billing administration and marketplace operations. When a customer supplies datasets, instructions or other personal data for a project, OpenTrain may act as that customer’s processor and follows the customer’s instructions and applicable data-processing terms. A business customer may request a Data Processing Addendum through the privacy contact.

Customers may provide datasets, instructions, prompts, files and related materials for freelancers or vendors to perform a project. We process that customer content to run the project, support collaboration and quality assurance, provide managed services, protect the Services and comply with law. Access is limited to authorized customer users, assigned freelancers or vendors, and OpenTrain personnel or contractors who need it for those purposes. Customers must have the rights and legal bases needed to provide the content and must give lawful instructions. Freelancers and vendors must use customer content only for the project, follow confidentiality duties and comply with communicated security requirements.

9. Payments and tax

We collect billing and invoice details, payment-method details such as the last four digits and transaction identifiers, payout-account status, tax forms and related compliance documents, transaction history, fees, refunds, chargebacks and payment status. Payment processors and financial partners may provide payment, payout, fraud and compliance signals and limited transaction details needed for accounting and support.

We process employer and client payments, facilitate freelancer and vendor payouts, calculate and charge fees, handle refunds, disputes and chargebacks, and keep accounting, audit and tax records. Stripe payment provisions in the Terms remain in force.

10. Marketplace sharing

The Services facilitate connections and transactions among users. Depending on role and settings, employers and clients may see freelancer and vendor profile information, applications, proposals, screening responses, interview information, communications, work product, reviews and marketplace history. Freelancers and vendors may see job posts, project descriptions, employer profile information, communications, task instructions, feedback and engagement payment or project status, but not full payment credentials. Some profile elements or marketplace content may be public or searchable depending on settings. Do not post information you do not want shared.

11. California Privacy Notice

This notice applies to California residents to the extent the CCPA/CPRA applies. In the last 12 months, we may have collected identifiers; California financial information; commercial information; internet or electronic-network activity; geolocation; professional or employment information; education information; and inferences. Sensitive personal information may include login credentials, government identifiers used for verification or tax, and financial-account details usually handled by payment processors. We use and disclose it only as permitted by law and as reasonably necessary to provide the Services.

We collect this information from you, your device and browser, service providers, other users and public or professional sources where permitted. We use it for the purposes in this Privacy Policy, including providing and securing the Services, payments, safety, support, analytics, advertising where consented, and legal compliance. We may disclose it to marketplace users, service providers and contractors, professional advisors, government entities when required, and parties involved in a business transfer. When you choose a buyer introduction, we send the selected buyer the company details, your full name and your work email, and OpenTrain may receive a referral reward from that buyer.

We do not sell personal information for money as a standalone product. Deepgram’s opt-out has been on since October 20, 2026, so current requests are not used for model improvement. Before that date, Deepgram may have received interview audio and used it to improve speech models in exchange for a 50% discount. That historical use may be a sale or sharing under CCPA/CPRA or another privacy law. LinkedIn server-side conversion measurement may receive a first-party ad click identifier, conversion time and opaque event ID after analytics/attribution consent; no name, email address, IP address or job data is sent. This may also be subject to sale or sharing rules. Use the Do Not Sell or Share My Personal Information section and Global Privacy Control where applicable.

California residents may request to know, delete, correct, opt out of sale or sharing, limit sensitive-personal-information use where applicable, and receive equal treatment. Submit a request to support@opentrain.ai with the subject “Privacy request”; we may verify identity before responding. An authorized agent may submit a request.