Privacy Policy

Last updated: November 17, 2026

Service provider: JEAR TECNICA S.A.

NIF: A28847549

Registered office: Paseo de la Castellana 259, 28046 Madrid, Spain

Contact: hello [at] picacho [dot] ai

See our Content Policy for the full rules on acceptable content, including strict requirements around real-people likeness and a zero-tolerance policy for anything involving minors. Content Policy

This Privacy Policy explains what information Picacho ("we", "us") collects when you use the Picacho website and app, how we use it, and the choices you have.

Who is responsible for your data

The data controller for Picacho is JEAR TECNICA S.A. (NIF A28847549), Paseo de la Castellana 259, 28046 Madrid, Spain. You can reach us using the contact details shown at the top of this page.

If you believe your data has been handled improperly, you can also lodge a complaint with the Spanish data protection authority (AEPD, aepd.es) or your local supervisory authority in the EU.

Information we collect

Account information: your email address and password (handled by our authentication provider), and optionally a username, company, and self-reported gender.

If you sign up or log in with Google or Facebook, that service tells us your name, your email address and your profile picture, and nothing more (not your contacts or your posts). We use them only to create your account and to recognise you when you log in. The language of the site or of your browser when you sign up is kept with your account, so our emails reach you in it.

Character and content data: the character profiles you create (traits, reference images), the prompts you write, the video clips you upload to Recast or Smash Cut (with their sound) and the pictures you add to a take, the images and videos generated for you, your notes, and your project organization.

Usage data: pages you visit and when, plus (once the app is deployed to production) an approximate country derived from your IP address, used to understand product usage. On the "Make your first picture" screen we also note which way you started (your own photo, Eva, or a character you described) and how far you got, as the name of each step and its time: never your words or pictures.

Error reports: if the app hits a fault while you are signed in, we record an automatic report on your account — what went wrong, the page you were on, your browser and device, the browser's own identifier (user agent), and the technical trace, trimmed to 1,000 characters. Administrators read these to fix the fault, and they are deleted with your account. Nothing is recorded for signed-out visitors.

Safety records: when our automatic safety check refuses a request, we record which check refused it, the reason, the time and a fingerprint (a hash) of the request, never its words. We also record any step we take on your account because of refusals (a warning, a pause, a review or a suspension) with its date and reason. When you confirm that you have the rights to a file you upload (on Generate or in Picacho Light), we keep that confirmation with its date, the wording you agreed to, where you gave it, your language and a scrambled (hashed) form of your internet address.

Cookies: small identifiers used to remember your theme and language, and — for logged-out visitors — an anonymous visitor ID for basic traffic analytics.

Payment information: payment details are collected and processed directly by our payment processor, Stripe. We never see or store your full card number — we keep only a reference to your Stripe customer and subscription so we can manage your plan.

How we use your information

To provide and operate the service — running the generation pipeline, storing your content, and keeping you signed in.

To maintain and improve reliability, including aggregate usage statistics about which features are used.

To communicate with you about your account.

If you stop using Picacho for a while after making something, if your first render doesn't work, or if you cancel a plan, we may email you once or twice to offer help, never more than one email in three days. If you sign up and haven't made anything yet, we may email you once or twice in your first days with a ready-made scene to try; those emails use set wording and only your first name and your newest character's name, and nothing is sent to an AI provider to write them. To write it, we send Anthropic (Claude, in the United States) what's relevant from your account: your first name, your characters' names, what you made recently with its date and match score, why a render didn't work, and your plan and credits. If you reply, Aly, Picacho's AI assistant, may answer, and says so; the team sees every reply. We keep your browser's time zone and the language you use the site in, so the email arrives in your morning and in your language. Each email has a link to stop them, and they stop on their own when you come back.

To keep Picacho safe and enforce our Terms of Service and Content Policy. Every request is checked automatically before anything is made. Refusals of your own words count over 30 days: at 5 we email you a warning, at 10 sending is paused for 24 hours and we email you why, and at 15 a person on our team reviews the account. Only a person can suspend an account, and we email you the reason. The warning and the pause are automatic; you can always ask a person to look again by replying to the email or writing to hello@picacho.ai. We do this to perform our contract with you and in our legitimate interest in keeping the service lawful and safe and meeting our AI providers' rules (GDPR Article 6(1)(b) and (f)).

AI-generated content

Prompts you submit, and the clips and pictures you use in a take, are sent to third-party AI providers to draft, review, and generate your content. These providers process your prompt text — and, for image generation, a reference photo if you've added one; for video, the clips and pictures the take uses — to produce a result. We don't control how these providers otherwise handle data beyond the terms of our agreements with them. One of them, Higgsfield, also keeps and uses what it receives for its own purposes; it only gets Full and Standard takes in Smash Cut and in Recast's Into the clip and Swap things that you explicitly agreed to send it (see its section).

When you use a clip in Recast or Smash Cut (one you upload or one of your takes), the clip and its sound are sent to Google (the Gemini API) to read what is in it before anything is made: who is in shot, where it cuts and whether anyone talks. If Google can't read it, still frames from the clip are sent to OpenAI for the same read instead. They use it only to answer that read.

Generated images and videos are stored in our cloud storage and are only accessible to your account, to anyone you share them with, and to Picacho administrators for support and safety purposes. If copying a video to our storage fails, it plays from the provider's own link instead, except a take made by Higgsfield, which never does (see its section). The provider that made a picture or video may keep its own copy for a time under its terms; Higgsfield's is described in its section.

Community: what other members see

Community is a wall inside Picacho that only signed-in members can open. Nothing you make appears there unless you post it yourself, one take at a time. A post shows the picture or video, the name you chose, your caption, the character's name, the result of the face check in words and, only if you choose to show it, the prompt. Members see how many hearts and views a post has; they don't see who gave a heart.

Your profile, at picacho.ai/app/community/ followed by @ and your name, shows members the name you chose, the month you joined, your posts, how often other members made a take from your scenes, and a top-three place in the weekly Casting Call. It doesn't show your email address or your real name. A name that was made from your email address when you signed up is not shown to anyone: your posts say "Picacho member" and you have no profile until you choose a name or keep that one. You can hide your profile from its menu. A post whose character you added with another person's permission stays off your profile. If you choose a picture or write a line for your profile, members see those too. The line is read by an automatic check (OpenAI) before it is saved. A picture you upload is cut square, loses the file's hidden details (such as where it was taken), and is read by our automatic safety check (OpenAI and Anthropic) before anyone sees it. We keep it until you replace it, remove it or delete your account.

You can take a post down at any time, and it leaves the wall and your profile at once. Deleting the take or your account deletes its posts and the hearts on them. You can report or block a member; they aren't told who reported or blocked them. We keep a report with the account that sent it so that a person can review it.

If we hide one of your posts, we email you what we hid and why, and you can reply to ask a person to look again.

Voice and the assistant

If you use a microphone feature (the mic in the composer, or talking to Aly, the personal assistant), your recording is sent to our AI provider, OpenAI, to be turned into text. Picacho does not store the recording; only the resulting text is kept, as part of your prompt or conversation. When the assistant answers out loud, its reply is sent to our voice provider, ElevenLabs, to be turned into speech, directly or through fal; if that is unavailable, OpenAI does it instead. The microphone is used only after you tap a mic button and allow access.

When you use Aly's live voice (it starts when you press Talk, unless you switch it off in Settings), your voice is streamed live to OpenAI, whose speech model listens and speaks for her during the call. Picacho does not store the audio of the call; the words said in it are kept as part of your conversation with Aly, and the length of each call is recorded with your assistant usage.

While you talk to the assistant, the words it heard are also checked by OpenAI, the provider that turns your recording into text, to tell whether they were said to it; anything that wasn't (a video playing, someone else in the room) is dropped and not kept. To spell names right, the transcription request also names the assistant (Aly, or the name you gave it) and your characters.

The personal assistant (Aly), on Elite accounts and on accounts we give it to, keeps your conversation with it and the short notes it writes about how you work, so it can pick up where you left off. You can read, edit and delete every note, and clear the conversation, from the assistant itself. Both are deleted with your account.

Aly's chat page (Aly in the menu) keeps each of your chats, the files you attach to them (PDFs, pictures, Word files, spreadsheets and text files) and the documents Aly writes there, so you can come back to them. What you write and attach is sent to the AI provider that answers it: OpenAI (GPT-6 Luna) by default; Anthropic (Claude) when you pick Claude or use Think harder on Luna; OpenAI (GPT) or Google (Gemini) when you pick one of them; and Anthropic, OpenAI and Google together when you use Ask all three. They use it only to answer. Each new chat is named by OpenAI (GPT-6 Luna) from the beginning of its first message and answer, whichever provider answered. When Aly searches the web for an answer, the search is run by the provider answering: OpenAI when Luna answers, Anthropic when Claude does. With </> Code on, Claude also runs code for you on Anthropic's computers: in a sandbox kept for each chat and, only when you press Start on a job card, on a cloud computer that can install software. What you ask and the files involved go to Anthropic only to do that, and the files it makes are kept in the chat's computer panel. Deleting a chat deletes its messages, documents, attached files and the files made there; files you attached inside a project stay with the project. Everything is deleted with your account.

Press Tour: your products, your ads and the accounts you connect

If you use Press Tour, we keep the product cards you make: the product's page address, the photos you pick or upload, the words on its label that you confirm, its logo box and colours, and your answer that you may advertise it. We also keep your brand kit and the ads you make: their plan, stills, filmed shots and finished files.

To read a product and to check each still and each filmed shot against its card, pictures of the product and frames of your ad are sent to Google: Cloud Vision reads the words on a label, and the Gemini API compares the product. They are used only to answer that check. We keep a record of every check (what was read and how it scored, without the picture) to measure how well the check works. The stills and shots themselves are made by our AI generation providers, like Picacho's other pictures and videos.

If you connect a social account (X, TikTok, Instagram or Threads) to post from Press Tour, we keep its name and handle and the keys the network gives us to post for you. The keys are encrypted before they are stored. We use them only to post what you check and send on the press line, when you choose, with the network's own AI label turned on; nothing is ever posted on our own. Each post is kept with a record of exactly what you approved and when, and a scrambled form of the internet address it was sent from. The networks you post to receive the posts you send, and their own privacy policies apply to them from then on.

If you use Clippings in Press Tour, we read your own posts to show you which of them work. With a connected Instagram or TikTok account and your permission, we read your newest videos, their captions and their views, likes and comments. To read what is said in a video (an Instagram video, or a video you add yourself and confirm you made), its sound is sent to OpenAI (Whisper) to turn it into words, and the words and captions are sent to OpenAI (GPT-6 Luna) to sort them into parts and name each video's format; they use them only to answer. We keep the words, the counts and the results, and an added video's cover picture; the file you add is deleted as soon as its words are read. We never read anyone else's account, and your posts are never combined with other people's. Disconnecting an account deletes everything read from it; deleting your Picacho account deletes all of it.

When you disconnect a social account, we delete its keys, cancel its posts that haven't gone out, and ask X or TikTok to end our access (Instagram and Threads don't offer that; you can also remove Picacho in their settings). Deleting your Picacho account deletes the keys and the records of your posts. If a network doesn't answer when we ask it to end our access, we keep that one key, encrypted, for up to a week only to ask again, then delete it. If you ask us to tell you when posting to a network opens, we keep that and email you once.

If you connect an app such as Claude or ChatGPT to your Picacho account, we keep a record of the connection: which app, what you allowed it to do, and when it was connected and last used. The app gets a key that works only for your account and only for what you allowed; we store only a scrambled form of it. The app can see your characters, your renders and your credits. A picture it makes for you spends your credits just as it would on Picacho; a video is only set up there and starts when you press Send on Picacho. What the app reads goes to the company behind it (for example Anthropic for Claude or OpenAI for ChatGPT) under its own terms, because you asked it to. An ad's paid steps start only when you tap Picacho's own card inside the app. You can disconnect an app at any time in Settings › Security › Connected apps, which ends its access at once; deleting your account ends it too.

Facial information (face verification)

Some video models refuse photos of real people. If a character is you, you can choose to verify your face so those models can use your photos. It is optional, it is only offered for your own face, and saying no does not affect anything else in Picacho.

What happens: our provider BytePlus (BytePlus Pte. Ltd., Singapore), which runs the Seedance video models, records a short live face check on its own page to confirm that a real person is present (not a photo, a replay or a mask). After a successful check, BytePlus keeps one reference image of your face. Up to three photos of the character you choose are then sent to BytePlus and compared with that reference image to confirm they show the same person; photos that do not match are not used. The facial information involved is the images and video from the check and the facial features extracted from them.

Why: only so the video models accept your own photos in videos you create from your own account. Your verified face is never used for anyone else's videos, for profiling or for advertising, and is not shared beyond BytePlus, which processes it only on our instructions.

Consent and records: we ask for your explicit consent before the check starts, and we keep a record of when you consented and to which version of this notice.

Where and for how long: BytePlus stores your facial information in its Southeast Asia region, outside the European Economic Area. It is kept until you remove your verification on the character page or delete your account. Removing your verification, or deleting your account, deletes it at BytePlus; if BytePlus cannot be reached at that moment, we retry every day until it is done.

Your rights: you can withdraw your consent at any time on the character page, which stops the processing at once. To access, correct or delete your facial information, or for any other request about it, contact us at the support email shown in Settings — requests come to us, not to BytePlus.

Character match check (face scoring)

When you make a picture or video of a saved character, Picacho checks whether the face stayed the same: OpenAI's vision model compares the new picture, or frames of the video, with up to three of the character's photos and returns a score from 0 to 100 and a one-line note. The score is shown with the result.

Whose faces are compared: characters you told us show no real person (drawn, 3D or made by AI) are always checked. A character you told us is you is checked only if you tick the separate face-matching box when you save it, or later on the character's page, and only for the photos you agreed for; untick it there at any time and the check stops for everything you make afterwards. Characters you told us show another person with their permission, and characters we have no answer for, are never checked.

What is kept: the score, the one-line note and the version of the check, stored with the record of the picture or video and erased when you delete your account. We never make or keep a face template. We also keep a record of when you ticked or unticked the face-matching box, and for which photos, until you delete your account.

OpenAI receives the pictures only to answer the check. It is based in the United States, so they leave the European Economic Area, and it may keep them for up to 30 days to check for abuse before deleting them. It does not use them to train its models.

For a character that is you, the check runs only with your explicit consent, which you can withdraw at any time without affecting anything else. For a made-up character, it is part of the service you ask for when you make a picture or video.

Full and Standard takes made by Higgsfield (Smash Cut, and Recast's Into the clip and Swap things)

In Smash Cut, and in Recast's Into the clip and Swap things, takes made at Full or Standard quality are made by Higgsfield, Inc. (535 Mission St, 14th Floor, San Francisco, CA 94105, United States) with an AI model it provides, through its API. Higgsfield makes a take only where the page names Higgsfield and asks for your tick. No other take is sent to Higgsfield: not Full takes in Recast's Restage it, Photo to life or Restyle everything, and not any take made with Kling 4 or at Lighter quality.

For each such take we send Higgsfield three things: the stretch of the clip you chose, as a copy without its sound and without the file's hidden details (such as where and on what device it was filmed); the pictures in the take, meaning your character's photos and any pictures you added; and the words for it, meaning the description we write, which never uses your character's name, and anything you wrote yourself, as you wrote it. Higgsfield fetches the files from links that work for a day. Those links contain your account's internal number and, for your character's photos, each photo's file name (for a photo you uploaded, based on the name it had on your device). The number doesn't tell Higgsfield who you are, but it does let it see which takes come from the same account. The requests come from our account, not yours: we add none of your account details, such as your name, your email or your payment details. Anything you write yourself, and your photos' file names, go as they are, so leave names out of them. Higgsfield sends back the finished video, which we copy to our own storage as soon as it is ready, and we put the sound back on from our own copy of the clip. We delete that soundless copy from our storage once the take is finished or has failed.

To make the take, Higgsfield's systems may analyse the faces in what it receives. Higgsfield says it uses those measurements only to make what was asked for and does not keep them (its privacy policy, section 1.6), and that it does not use the faces it receives to identify anyone (its Terms of Use, section 5.3). That promise covers only the measurements: the clip, the pictures and the take themselves are kept, as described below.

Higgsfield's terms (sections 4.3 and 4.4) let it keep what it receives and the videos it makes, and use them to run its service, including checking content, and to train and improve its own and its affiliates' AI models. Its service providers may handle them for it. On 13 October 2026 we switched on the setting in our Higgsfield account that excludes from model training what our requests generate. Higgsfield's documentation says that setting covers the outputs, meaning the finished takes; it doesn't say that it covers the clips, pictures and words we send. So unless Higgsfield confirms otherwise in writing (we have asked it to), it may use what we send to train and improve its AI models, and its terms say that a model already trained on something isn't untrained when that is deleted (section 16.5). It may use them in its own marketing only if they are public on its service or if we agree, and we don't.

Higgsfield has accepted no data processing terms with us, so it is not bound to use what we send only to make your take. It also uses it for its own purposes: its privacy policy (higgsfield.ai/privacy-policy) calls Higgsfield the controller of what it receives (section 12.3) and names its own legitimate interests, not your consent, as its legal basis for running and protecting its service, such as checking content and keeping the service safe (section 12.7).

Higgsfield is in the United States. Its privacy policy says it also has a location in Kazakhstan and shares data with service providers around the world, including other AI model and computing companies. Its list of them (trust.higgsfield.ai) is now given only on request; when it was last published, it named companies in the United States, Singapore, India, Israel and the EU, and didn't say which of them work on these takes. No EU adequacy decision covers this transfer. The one for the United States covers only companies certified under the EU-U.S. Data Privacy Framework, and we found no Higgsfield entry on the official list when we last checked, on 4 November 2026. Kazakhstan, Singapore and India have no EU adequacy decision at all. We have also signed no standard contractual clauses or other GDPR safeguard with Higgsfield. So we send it only if you explicitly agree to this transfer (GDPR Article 49(1)(a)). This carries risks for you: the law of those countries may not protect your data as EU law does; there may be no independent data protection authority there for you to turn to, and you may not be able to enforce your GDPR rights against Higgsfield or its providers; and public authorities there may be able to obtain your data under their own laws.

So we ask first. Each time you pick a clip (in Smash Cut, an edit) for these takes, you tick a declaration about it: that it is yours to use, that the only real person in it and in the pictures is you, or nobody real, that no child appears in them, and that you accept the rules for these takes in our Terms. For an edit from Picacho's own library, we give the edit's rights and answer for who appears in it, and your declaration covers your character, its pictures and the words you add; an edit a child appears in is offered only with Kling 4. That declaration covers the takes you then make from that clip, and you are asked for it again when you pick a clip again or change the quality from Kling 4 to Full or Standard. In a separate tick, given once and remembered as described below, you explicitly agree to the transfer and to Higgsfield's own use described above: keeping what we send and the takes, using them to run its service, and using what we send to train and improve its AI models unless it confirms otherwise in writing. That explicit consent is our legal basis for sending anything to Higgsfield (GDPR Articles 6(1)(a) and 49(1)(a)). Full and Standard here are only made if you agree. Smash Cut and Recast start you on Kling 4, and a recipe made at Full or Standard runs at that quality only after you tick. You can always make the take with Kling 4 instead: Kling 4 takes are never sent to Higgsfield, though they are made through fal, in the United States, and the companies whose models fal runs (such as Kling), so they leave the European Economic Area too.

For a character, "the only real person is you" means a character you told us is you, or one you told us shows no real person, for the photos it has now. Characters you told us show another person with their permission, and characters whose current photos we have no answer for, are never sent to Higgsfield.

With each take we keep a record of what you confirmed and agreed to, the version of the wording you saw and the language it was shown in. Your agreement is remembered for your account only in the browser or app where you ticked it, until you untick it there, so untick it in each one you use. Unticking stops anything new being sent to Higgsfield from there. A take already started still finishes, and what was already sent stays with Higgsfield, as described below.

If you appear in a clip or picture someone else sent for one of these takes, that broke our rules; we rely on the sender's word and can't check every clip. That Picacho member then gave us your image, and your voice if it is heard in the clip, without your consent; we sent your image to Higgsfield as described above (without the sound), and we keep the clip or picture with their take. Tell us using the contact details at the top of this page, with the clip or a description of it; if you have seen the take itself, you can also report it at picacho.ai/report. We will remove it from Picacho (kept unserved as evidence, as described under "Reports about content"), ask Higgsfield to delete its copy and not to use it, and tell you what Higgsfield answers. You have the rights described under "Your rights", including to object and to complain to the AEPD.

We keep the clip, the take and the record of your agreement as described under "Data retention". At Higgsfield, each take's request, meaning what we sent and the take it made, is kept for at most an hour after Higgsfield finishes it: we ask for that, the shortest time it offers, with every request. As soon as the take ends, whether it was stored, failed or was stopped, we ask Higgsfield to delete the request, which removes its files, the words and the settings and keeps only the request's id, model, status and times and its billing record. If copying the finished video to our storage fails, we try again while Higgsfield still has it; if it still can't be copied, the take fails, and it is never shown from Higgsfield's own link. Higgsfield's documentation says that copies of what it receives that it prepares for reuse are deleted within seven days, and that a request run by another company's model is also subject to that company's data policy. Its privacy policy says it keeps what it receives for as long as it needs it, and that content removed there can stay on its servers for 30 days and in its backups after that (sections 3, 5.9 and 13). Its terms also let content be shown to others where its settings make it visible, including by default; its API gives us no such setting, and we don't know whether it ever shows takes made through it.

You can withdraw your agreement at any time by unticking the box, and you have all the rights described under "Your rights". For what Higgsfield holds, write to us at the support email shown in Settings or using the contact details at the top of this page. Because the requests come from our account, Higgsfield can't find your takes from your name, so we send Higgsfield your request (to see what it holds, to delete it, or to object to its use) with your takes' references, and tell you its answer. You can also contact Higgsfield yourself at privacy@higgsfield.ai, or through its EU representative, DataRep (higgsfield@datarep.com; 72 rue de Lessard, 76100 Rouen, France). Deleting a take, or your account, also asks Higgsfield to delete what was sent for those takes and the videos it made, if it still has them. What Higgsfield has already used to train its models stays in them, as described above.

Reports about content

If you report content at picacho.ai/report, we keep what you send: the category, the links, your explanation, your name, email and typed signature, any copy you upload, and a scrambled (hashed) form of the address you sent it from, used only to stop floods of reports. We use it only to handle your report and to show what we decided, and we email you to confirm the report and to tell you the decision.

We don't give your name or email to the person whose content you reported, unless the law requires it. A report that is never confirmed is deleted after 7 days, with any copy you uploaded. A decided report is kept as a record of what we did for at most 3 years.

To find copies of reported content, we keep a fingerprint of every picture and video made on Picacho: a short code that describes the picture's light and dark as a whole. It is not a face template and can't be turned back into the picture. It is stored with the render's record and deleted when the account is deleted. Content removed after a report is kept, unserved, for 90 days as evidence and then deleted; content reported as involving a child may be kept longer so it can be handed to the authorities.

Cookies & analytics

We use cookies for essential functionality — staying signed in, remembering your theme and language — and for basic, privacy-conscious analytics, such as page views and approximate country, so we can understand how the product is used. We do not sell this data.

Sharing your information

We share data with the service providers who help us run Picacho: our database, authentication, and file storage provider; our hosting provider; our AI generation providers (one of them, Higgsfield, also uses what it receives for its own purposes, and gets only Full and Standard takes you explicitly agreed to send it; see its section); and, once enabled, our payment processor. We do not sell your personal information to third parties.

With each request, OpenAI and BytePlus (BytePlus Pte. Ltd., Singapore, which runs the Seedance video models) also receive a scrambled code made from your account, so that if someone misuses their models through Picacho they can act on that one account instead of on Picacho as a whole. The code is the same every time for you, contains no name or email address, and only we hold the key that links it to your account. Each provider gives every request a task number, which we keep with the picture or video it made, so a request can be traced if a provider asks.

If an AI provider whose rules require it (BytePlus's rules for platforms like ours do) asks us for records about misuse of its models through Picacho, we may give it what is needed about the accounts concerned and nothing more: the code above, what was made or refused and when, the safety steps we took, and the upload-rights confirmations. We give information to courts and authorities when the law requires it.

Where these providers are outside the EU: OpenAI's terms place a European customer like Picacho with OpenAI Ireland Ltd., in the EU, and when OpenAI moves data to the United States or elsewhere it does so under the European Commission's standard contractual clauses or an EU adequacy decision (OpenAI Data Processing Addendum, section 4.1). BytePlus is in Singapore, which has no EU adequacy decision; its Data Processing Addendum, which is part of our agreement with it, includes the European Commission's standard contractual clauses (Decision 2021/914), with Picacho as the exporter and BytePlus as the importer. You can ask us for a copy of the clauses that apply to your data.

Data retention

We keep your account and content for as long as your account is active. If you delete your account from Settings, your profile, characters, generations, notes, and projects are permanently deleted. Anonymized traffic records that can't be linked back to you may be retained for analytics.

For 120 days after you delete your account, we keep a short record of it with no name in it: when you joined and left, your plan, how many pictures and videos you made, and when we emailed you and whether those emails arrived. Your email address is kept only in a scrambled (hashed) form, so that if you write to us after leaving we can tell it was your account. After 120 days the record is deleted.

If your account had requests refused by our safety check, any safety step or a suspension, deleting it leaves one locked record, kept for 3 years from the deletion and then deleted: the scrambled code described in "Sharing your information" (never your name, email address, requests, pictures or videos), the safety steps with their dates and reasons, how many requests were refused and for what reason, and which upload-rights wording you confirmed and when. We keep it only to establish, exercise or defend legal claims, for example a claim by an AI provider about misuse of its models through the account (GDPR Article 17(3)(e)), and it is blocked as Spanish law requires (LOPDGDD Article 32): it is used for nothing else, and only a court, an authority or a legal claim leads us to open it. Accounts with no refusals, steps or suspension leave no such record.

Records of what you paid for — the plan, credit pack or top-up, the date, the amount and currency, whether it was refunded, and the payment provider's reference — are kept for six full years after the year of each payment, as Spanish accounting and tax law requires, even if you delete your account; once the account is deleted they are no longer linked to it. Our payment processor, Stripe, keeps the invoices and the billing details on them for the same reason, and Google keeps its own records of Google Play purchases.

At Higgsfield, each Full or Standard take's request, meaning what we sent and the take it made, is kept for at most an hour after Higgsfield finishes it, and we ask Higgsfield to delete it as soon as each take ends; deleting a take or your account asks it again. Copies Higgsfield prepares for reuse can stay there up to seven days, and its own privacy policy lets it keep content longer; what Higgsfield has already used to train its models stays in them. Its section describes all three.

Your rights

You can view and edit your account details, change your password, and permanently delete your account at any time from Settings. Picacho is operated from Spain, so the EU General Data Protection Regulation (GDPR) applies to how we process your personal data: you have the rights of access, rectification, erasure, restriction, portability, and objection. To exercise any of them, contact us at the support email shown in Settings or, if you don't have an account, using the contact details at the top of this page, and we'll respond within the timeframes the GDPR requires. For takes made by Higgsfield, we also pass your request on to Higgsfield, as described in its section. You also have the right to lodge a complaint with the Spanish data protection authority (AEPD) or the supervisory authority where you live.

The safety warning and the 24-hour pause are taken automatically. For any step taken on your account, you can ask a person to review it, give your point of view and contest it, by replying to the email about it or writing to hello@picacho.ai.

Children's privacy

Picacho is not directed at, and may not be used by, anyone under 18, and we do not knowingly allow accounts for anyone under 18. Clips and pictures sent for a Full or Standard take in Smash Cut or in Recast's Into the clip and Swap things must not show a child. If we learn that one did, we ask Higgsfield to delete its copy and remove it from Picacho, keeping it unserved only as evidence, as described under "Reports about content". Anyone, including a parent, can tell us about content showing a child at picacho.ai/report or using the contact details at the top of this page.

Security

We use industry-standard practices — encrypted connections, access controls, and row-level security on our database — to protect your information. No method of storage or transmission is 100% secure.

We keep a written plan for security incidents. If a breach of personal data is likely to put your rights at risk, we tell the Spanish data protection authority (AEPD) within 72 hours of learning of it, and if it is likely to put you at high risk we tell you without undue delay.

Changes to this policy

We may update this policy as the product evolves. We'll update the date above whenever we do.

Contact us

Questions about this policy? Reach us at the support email listed in Settings, or using the contact details at the top of this page.