Privacy Policy

Toast is an AI-powered video and audio editor, operated by Arnanda Technologies Private Limited ("we", "us", "our"), a company registered in India and the data controller for the information described below. We are committed to protecting your privacy, especially when it comes to your content.

When you edit your own video, it never leaves your device. Every frame is decoded, edited and exported in your browser, with tools that run on your device. Video files live in your browser's Origin Private File System (OPFS) and are never uploaded to Toast's servers or to any third party, unless you choose to share a finished video on Toast (below). Page recordings (section 3) can hold small pictures taken from it, such as project covers.

Audio is the deliberate exception. Turning speech into words with timings cannot happen on your device, so we extract the audio track and send it to the services described below (section 4 names them) — and for a video built from a voiceover, the file you upload is audio, so that file is what gets sent. Section 3 lists everything that leaves and why.

Property videos work differently. If you make videos of a property (a home, an office, a project), you upload its shoot: the videos, photos and audio. They are stored in Toast's private storage so Toast can read them and make your videos on its servers. Toast's own servers also make smaller copies of each file (a preview video, stills, the sound track, a display copy of each photo) and keep them beside it. Nobody else can see them. When you ask Toast to read the property, stills of each file go to an AI model and each sound track to a speech-to-text service; when you ask for a video, what is said to the camera in the shoot goes to an AI model to plan it (section 3). When you ask for the finished video, Toast's own servers make it from your original files and keep it with the property, and stills of it, and pictures of it with its words and cards on them, go to an AI model to check it before you see it. We keep the files while the property is in your account; deleting the property, one of its files, or your account deletes them.

Screen recordings you make an animated video from. Your screen recordings stay on your device. When you choose to make an animated video from one or more of them, each of them is uploaded to Toast's private storage. Toast keeps each recording while a video made from it exists and deletes it with the last such video or your account. Toast's own servers make a smaller copy, pictures and the sound of each; these are stored with it and deleted with it, and are sent to no one by themselves.

Videos you share on Toast. If you press Share on Toast on a finished video, that video (the file you exported, never your original footage or your project) and a still from it are uploaded to Toast's private storage, with the title you type and the credits for any music, stock photos or footage in it. Toast's own servers make a copy without location, device or time details. Nobody else can see it until a person at Toast has watched it and approved it. Then that copy, the still, the title and the credits are public on Toast's "Made with Toast" pages, where anyone can watch them. A video we do not approve is deleted. You can unshare it at any time, and deleting the video's project or your account unshares it too; either way the public copy is deleted within a day. Before you share, you confirm that everyone who appears or speaks in it agreed to it being public, and that no child appears in it unless you are their parent or guardian.

Reports about a video. If you report a problem with a video, we keep your report, including the moments you mark and any note you write, so we can look into it. It is deleted with your account.

Sharing a video to get its cost back. If you share one of your videos to get its cost back, we keep the link to your post, the account name you give us and which video it was, so a person at Toast can check the post. The screenshot you send is kept privately until we decide, then deleted. The rest is deleted when your account is deleted. We never download your post or send it to anyone. A person at Toast opens the link like any viewer, and if you say yes, we show the link on Toast's 'Made with Toast' showcase.

Event videos work the same way. If you make videos of an event (a function, a launch, a talk), you upload the clips filmed at it: the videos, photos and audio. They are stored in Toast's private storage so Toast can make your videos on its servers, and Toast's own servers make the same smaller copies of each file and keep them beside it. Nobody else can see them. When you ask Toast to read the event, stills of each file and what its sound holds (who speaks when, where there is music, applause or laughter) go to an AI model, and each sound track to a speech-to-text service (section 3). A clip you add to two of your events is stored once. We keep the files while an event in your account uses them; deleting the event, one of its files, or your account deletes them. Toast never recognises the faces of the people filmed.

• Audio for transcription and noise removal: To transcribe your recording and optionally remove background noise, we extract the audio track and send it to a speech-to-text service and a noise-removal service. When you edit your own video, only the audio is transmitted — never the video. For a story video made from a voiceover you upload, the box “Clean this audio first” is ticked unless you untick it: that audio then goes to the noise-removal service as well, and Toast takes out long pauses, filler sounds and repeated lines on your device.
• Instructions you say instead of typing: In a box where you tell Toast what you want, you can tap the microphone and say it. Your device records what you say; Toast sends that recording to a speech-to-text service once to write down its words and shows them to you in the box to correct. Nothing of the recording is kept, and neither are the words: only the request you then send, as with typed words.
• Your property shoot: When you make videos of a property, the files you upload (its videos, photos and audio, with their names, sizes and types) are stored in Toast's private storage, held by our storage provider, together with the details you type about the property (such as its price, size, licence number and client name). Your browser sends each file straight to that storage; it is not passed to any third party. Toast's own servers then check that each stored file is the one you sent and make smaller copies of it to work from (a preview video, stills, the sound track, a display copy of each photo). The copies are stored beside the original without its location, device or time details, and are deleted with it. Toast's own servers also mark where faces appear in those stills (a box around each face, never who it is: no face is recognised, matched or compared), so the words on your videos are placed around faces rather than over them. The boxes are stored with the copies, sent to no third party and deleted with them. Until you ask Toast to read the property, nothing of the shoot goes to any third party. Photos you upload of a property are also sent once to Toast's own graphics service, to work out their depth for gentle camera moves; no third party receives them.
• Before photos. If you upload photos of a space as it was before and mark them “before”, they are stored, prepared and deleted exactly like your other files. Toast's own worker finds which finished shot shows the same angle (numbers only, sent to no one), and you confirm each pair. Their stills go to the AI model with your other stills, through our AI routing service on your key, marked “before” so the reader can name the space, and the plan is told which rooms have one. The before-and-after reveal is cut by Toast's own worker. Toast never makes a “before” picture with AI.
• Your finished property videos: When you ask Toast to make a planned property video, Toast's own servers make it from your original files (no third party receives them for this) and the finished video is stored in Toast's private storage, without location, device or time details, so you can watch and download it. It is deleted with the property or your account. Its captions are drawn from the words said to the camera: those words go, a few seconds at a time, to Toast's own service that draws graphics (no third party), and the caption clips it draws are kept with the video and deleted with the property. Its room labels (“Master bedroom”) are drawn the same way: the names of the rooms it shows, as you left them, go to the same service (no third party), and the label clips are kept and deleted the same way. So each label suits the picture under it, Toast's own video service first looks at the frames where it will sit and measures how light and how busy each place is (nothing is kept and nothing leaves); the label then goes to the graphics service with its number in the video, how it is drawn and one colour taken from your footage for its accent line. A video may also name a detail Toast saw in a shot (“Fluted oak panels”) while it plays: the words Toast's reading wrote for that shot go to the same service (no third party), and the clips are kept and deleted the same way. A detail is drawn on a small paper tag in the same place on every shot or, when Toast knows where it is in the picture, on a tag beside it, measured the way the labels are (nothing is kept and nothing leaves); no colour is taken from your footage. Your brand kit (your business name, phone, WhatsApp number, theask you write in your own words, website, licence number and the logo you upload) is stored by Toast until you change it or delete your account, and is drawn on the end card of your property videos, its ask, its WhatsApp number or phone or website and name on a small card that asks part-way through when a video's plan has one, and its name, with the bedrooms, kind and locality you typed, on a short title over the first seconds when a video opens on its finest detail (or over the next shot when the title would cover that detail): those words and the logo go to the same service (no third party), never to an AI model, for these videos (an animated video you make in your brand sends them to an AI model, as described under animated videos below). People who watch a video answer you on your own WhatsApp or phone: Toast collects nothing from them, and the video carries no link. If your brand kit has colours, your property videos' end card, lead card and price card use them, drawn by Toast's own renderer; they are sent to no one else. Itsfacts card, near the start, again before the end card, or only one of the two as the video's plan picks, shows the facts you typed and chose to show (the building's name, the bedrooms and kind, the size, the locality, the price, an interior studio's budget and the time the job took, up to three lines of your own and the licence number): those words go to the same service (no third party), never to an AI model, and the clips are kept with the video and deleted with the property. When you choose toreveal the price later, the price, the licence number and the words you typed over the price are drawn once more, on their own, by the same service (no third party), never by an AI model. When you pick astatus for a property (just listed, open house, sold or a new price), a banner opens its videos: the status, the open house's day and time as you typed them and the facts it shows go to the same service (no third party), never to an AI model, and the status is stored with the property. When you pick a quote for a video, a line your client said to the camera is shown on screen while the music plays, exactly as they said it or shortened only in the version you approved (the “…” shows where words were left out): those words and the name you typed for who said it go to the same service (no third party), never to an AI model, and your picks are stored with the video and deleted with it. When Toast plans a property video, it also sends an AI model (through our AI routing service, on your key, which asks it not to keep or learn from what it is sent) one still from each shot in that video (the small stills it already took from your footage, never the video itself), each shot's room name and the features the reading saw, who the video is for, and the note and style lines you typed, once per video, to decide whether anything in a shot deserves a highlight, why, where it is in the picture and how to draw it, and where each of those features is in the still when it can be seen clearly, so its tag can be placed next to it. Its answer is stored with the video and deleted with it. Following those items through the shot, and measuring where each tag reads best, happen on Toast's own servers, and Toast's own renderer draws the highlight and the tags. You can remove any highlight, or add one by tapping an item. If you change a property video's words, labels, tags or titles (what they say, where they sit, when they show, or hiding one) or add your own, Toast stores your changes with the video and deletes them with it; your words go to the same service that draws your cards (no third party), and when Toast checks the finished video, pictures of it with your words on them are seen by an AI model (through our AI routing service, which asks it not to keep or learn from what it is sent). If you type proof lines in your brand kit (such as “500 homes delivered” or “Rated 4.8 by our clients”), they are stored with your brand kit and shown exactly as you typed them on your property videos, in the end card and once part-way through: they go to the same service that draws your cards (no third party), never to an AI model, and are deleted with your account. If you type a room's size or its finishes (such as '12 ft x 14 ft' or 'Italian marble'), we keep them with the property and Toast's own graphics service draws them on that room in your video; no third party receives them, and Toast never measures or guesses a size. If you type nearby places and their distances (such as 'Metro 800 m'), we keep them with the property and Toast's own graphics service draws them as a card in your video; Toast never looks them up and sends them to no one else. A video planned for an investor opens with a numbers card: the price, the expected rent, the rental yield and the price per area exactly as you typed them (Toast never works one out), with the licence number when you typed it, stored with the property's other facts; those words go to the same service (no third party), never to an AI model, and Toast checks when it makes the video that they are big enough to read on the screen the video is made for. A video planned for a tenant opens with a rent card: the rent, the deposit, the day it is free and the furnishing exactly as you typed them, stored with the property's other facts; those words go to the same service (no third party), never to an AI model, and the same check makes sure they can be read. On a broker's or developer's video the licence number goes only whileAdd RERA details is on (it starts on when you have typed one, and you can turn it off for a property or a video): off, no card carries it and it is not sent. Before you see a finished video, Toastchecks it: its own servers measure the picture and the sound (nothing leaves for that), and stills of the finished video (one small picture from the middle of each shot of a room, never the video itself), the room names and the sentences said over those shots are sent to an AI model, through our AI routing service, which asks it not to keep or learn from what it is sent, once per video, to say whether each shot shows the room it is named as and the words fit the pictures. The stills are not kept; what the check found is kept with the video. Toast then watches the finished video as a viewer would: small pictures from across the whole video (540 px wide, about one every second) with everything Toast drew on it, and the words shown and spoken in it with when each comes, and the room names, are sent to an AI model, through our AI routing service, which asks it not to keep or learn from what it is sent, to say whether every word can be read on a phone, sits by what it names, and whether the video is ready to post. Its sound is checked on Toast's own servers and is never sent. When it finds a problem, Toast fixes the words, or leaves out or replaces a shot with another of the same room, and makes the video again, at most twice. Each time it sends only the pictures that changed, with the AI model's own earlier scores and comments on that video. When Toast makes a video again to fix something, the same AI model is shown pictures of both makes, the same way, to say which is better; Toast keeps the earlier one if the new one is not better. The pictures are not kept; the scores and what was found are kept with the video. To make these checks better, the person who runs Toast may watch finished property videos and mark whether each is ready to post; that mark is kept with Toast's own notes, never shown to anyone else, and deleted with the video.
• Your event footage: When you make videos of an event, the files you upload (its videos, photos and audio, with their names, sizes and types) and the name you give the event are stored in Toast's private storage, held by our storage provider. Your browser sends each file straight to that storage. Toast's own servers then check that each stored file is the one you sent and make smaller copies of it to work from (a preview video, stills, the sound track, a display copy of each photo), stored beside the original without its location, device or time details and deleted with it. Toast's own servers also mark where faces appear in those stills (a box around each face, never who it is: no face is recognised, matched or compared), so the words on your videos are placed around faces rather than over them. The boxes are stored with the copies, sent to no third party and deleted with them. Until you ask Toast to read the event, nothing of an event's footage goes to any third party.
• Reading your property shoot: When you ask Toast to read a property, stills of every file (up to twelve per video, small copies, never the files themselves) are sent to AI models, through our AI routing service, which asks them not to keep or learn from what they are sent: one model says what each file shows and which files show the same room, another checks the shots for problems and rates how good each one looks, as a room and as a close-up of the work, so Toast can suggest your best shots first. Each file's length and whether it has speech go with them; when it reads, the transcript, the file names and the details you typed do not. Each file's sound track (the audio only) is sent to a speech-to-text service once to be transcribed, and the words it hears, with their times, are kept with the file. Reading again sends the stills again, but never a sound track already heard.
• Planning a property video: When you ask Toast to make a video of a property (for example a testimonial reel), the sentences said to the camerain the shoot (the words transcribed from the clips where someone talks to the camera, with how long each lasts) and the names and visible features of the rooms Toast found are sent to an AI model, through our AI routing service, which asks it not to keep or learn from what it is sent, once per video, to choose which sentences to use and what to show while each plays, and once more to check that the chosen sentences tell their story in an order that makes sense and that what is shown under each suits it (again if Toast plans it again). If instead you ask for a video in your own words (“just the kitchen”), what you typed goes to the same models the same way, with the room names and what each usable shot shows and how long it is. No picture or file name is sent. Of the details you typed about the property, only one kind is: when an interior studio's video is planned to a plan Toast offers (the reveal, in their words, room by room), the style or materials lines you typed for the facts card go with Toast's direction for that plan (how it opens, its order, its length, who it is for), so the close-ups match what you named, with what you typed about what the video is for if you did (“new villa clients in Kompally”); those plans are made by our most capable AI model. When you say who speaks in a take (the designer, the client or someone else), Toast sends that choice with the take's sentences to the AI model that plans the video. When a broker's or developer's video is planned for a buyer, an investor or a tenant (the walk-in, the view first, a quick look, the numbers first, the home then the numbers, the move-in), Toast's direction for that plan goes to that same model the same way, with what you typed about what the video is for if you did (“NRI investors”), and nothing else you typed: its numbers are drawn by Toast, never planned by a model. A budget, a time, a price or any other detail you typed is never sent. The plan (the chosen sentences or shots, and your request) is kept with the video.
• Reading your event footage: When you ask Toast to read an event, each file's sound track (the audio only) is sent to a speech-to-text service once to be transcribed, and the words it hears, with their times and which voice said them, are kept with the file (a clip in two of your events is heard once). Stills of every file (up to twelve per video, small copies, never the files themselves) are then sent to an AI model, through our AI routing service, which asks it not to keep or learn from what it is sent, with each file's length and what its sound holds: how many people speak and when, where there is music, applause or laughter. The model says what happens in each file, the moments worth keeping, where a child can be seen, and which people are featured, describing them by what they wear and do. The transcript's words, the file names and the event's name are not sent. Toast never recognises faces: the model is told to tell people apart only by clothing, place and role, never by their face, and to name no one; you say who is who. What it said is kept with the event, and so are the names and roles you type for the people in it, which are sent to no one. Reading again sends the stills again, but never a sound track already heard.
• Planning an event video: When you ask Toast to make a video of an event, the sentences said in its clips (the words transcribed from them, with how long each lasts and which featured person the reading thinks said it, shown as “P1”, “P2”), the featured people as those labels with how the reading described them (their role and what they wear), the moments it found, its one-line summary of the event, the length you chose and what you typed about the video, if anything, are sent to an AI model, through our AI routing service, which asks it not to keep or learn from what it is sent, once per video, to choose which lines and moments to use and in what order. No picture, file name or the event's name is sent, and the names and roles you typed for people are not sent: Toast puts them on screen itself. Unless you say the children seen are your family or you have their parents' agreement, no clip in which a child can be seen is shown. The plan (the chosen lines and moments, and what you typed) is kept with the video.
• Your finished event videos: When you ask Toast to make a planned event video, Toast's own servers make it from your original files (no third party) and keep it with the event. Toast's own renderer draws on it the captions (the words the video plays), the names and roles you typed for the people who speak in it, and its title; no third party receives them. The video, and the caption and name clips drawn for it, are deleted with the event or your account.
• Transcript text for AI editing: The transcript produced from your audio (text and timestamps, plus the editing instructions you provide) is sent to our AI language-model providers so Toast can suggest cuts, captions, and effects, and so it can plan the graphics, zooms and on-screen words that change every few seconds. That plan is made once for each shape you make the video in (for example wide and vertical), so the transcript is sent once per shape. For that plan your browser also sends the timing of each spoken word to Toast's own server, which uses it to time each graphic to the word it lands on; the word timings are not passed on to the AI providers. Nor, for that plan, is where your edit cuts: neither the times of the cuts nor which words fall either side of them. Choosing a transition between two shots is different: for that, the AI is shown each cut in your edit, with its time in the recording and the few words said either side of it, so the transition suits what is said there. If you enter a speaker name and role on a project, that name and role are sent with the transcript for the same plan, so a lower third can show them; leave them empty and no name is sent or shown. When the AI plans graphics for a talking-head video, it is also given a short caption for each picture already in that video (a licensed photo of a person or company the video names, or a picture Toast drew for it) so it can put them on picture cards. It never receives the pictures themselves. If you choose captions in English letters (for example Hindi written as “kya aapka”), the distinct words of your transcript, as a list without their order or timings, are sent to the same providers to be written that way; nothing is sent for this unless you choose it. If you choose captions and titles in another language, the caption lines of your transcript, in speaking order but without their timings, and the words of the AI's on-screen cards are sent to the same providers to be translated; the translation is kept on your device, and nothing is sent for this unless you choose a language other than the one spoken. If you put labels on people or things in your video and then ask the editing chat for a change, the words of your labels, when each one shows, and the time of a frame where you clicked something but have not labelled it yet are sent with the transcript to the same providers, so a request such as “put Chef Ana on the woman” can find it. Following the person or thing happens on your device: the picture, its outline and where it is on screen are never sent. This is text only — no video or audio frames are sent to these providers.
• Transcript text for planning a story video: For a video built from a voiceover, the same transcript is used for three further purposes, and all three send it to the same AI language-model providers. First, to work out who and what recurs in your script, so the same person or object can be drawn consistently. Second, to plan the picture for every moment of the video — which means the transcript is sentwith its sentence timings, because every scene boundary is derived from when a line is actually spoken. Third, to plan the graphics cards over those pictures — the same graphics plan described above, with the timing of each word again kept on Toast's own server and not passed to the providers. For that graphics plan, the providers are also given a short description of each picture drawn for your video (a few words about what it shows, written by Toast from your script) so a card can gather several of them; never the pictures themselves. Still text and timestamps; still no audio or video. When your script was a production document, both steps also send the notes you wrote in it about the pictures — the style note, the image notes and any cast descriptions — so that a place, a thing or a person is described in your words rather than guessed. Those notes are text you wrote, and they go to the same providers. What is kept with your project is what Toast writes from them — the descriptions of your cast and of each scene — not the document.
• The words on a graphics card: When Toast puts a designed card on screen — a statistic, a label, a title — the card is drawn on our own servers and sent back to your browser as a small transparent clip, which your browser lays over your footage. Your video is not involved: the renderer never receives a frame of it. What we send to that renderer is the text that appears on the card — usually a few words taken from your own script — along with which card design to use and how big to draw it. It goes to Toast's own render service and to no third party. We deliberately do not send where the card will sit or where your face is in the frame: your browser works that out itself, and the renderer has no use for it. The finished card is stored with your project (see section 4) so that re-editing your video does not redraw it.
• Your captions, when Toast draws them: Captions are drawn the same way as graphics cards — on Toast's own render service, as transparent clips your browser lays over your video. To draw them we send that renderer the words spoken in your finished video and when each one is said — the transcript of what your edit keeps, up to 15 seconds of it per clip, so across a whole video the whole kept transcript — with the caption style, its size and two colours your browser chose from the strip of the frame behind the captions. Your video is not sent, and neither is that frame. It goes to Toast's own render service and to no third party. The drawn caption clips are stored with your project like cards (see section 4), so they show your words; the list of words and timings is not kept with them.
• Pictures on a graphics card: Some card designs show a picture — a picture reveal, a phone or browser screen, a gallery. The picture is always one Toast already keeps for you: a picture Toast drew for your video, a licensed photo or logo Toast found for a person or company your script names, or a picture you put on a frame yourself. To draw the card we send that picture, made smaller to fit the card, to Toast's own render service and to no third party; the renderer fetches nothing itself. The finished card is stored with your project like any card (see section 4), so it shows the picture. A reference photo you upload of someone in your cast is never put on a card, and a frame of your video cannot be: your video never leaves your device.
• Pictures Toast drew, for camera moves: To move the camera through a picture Toast drew for your video — near things moving more than far ones — each such picture is sent once to Toast's own render service, which works out how far away each part of the picture is and sends back a grey “depth map”. It goes to no third party. Only pictures Toast drew are sent — never a photo you uploaded to a story video (a property's photos are the one exception, above), never a licensed photo or logo, and never a frame of your own video. The depth map is stored with the picture (see section 4) and deleted with it.
• One frame’s description, to improve it: On the review screen you can ask Toast to rewrite the description of a single picture. That sends the description and the one line of narration it illustrates to the same AI language-model providers — one sentence of your script, not the script — and the suggestion is handed back to you and stored nowhere. Toast also does this by itself while planning your pictures, for any description that asks for a map of particular places (image models draw the wrong places): the same two things go to the same providers, and the rewritten description replaces the planned one in your plan.
• A script document you ask Toast to read: If you write or load a script that is a production document — with a title block, image notes, music cues or stage directions in it — Toast sorts it into what should be spoken and what should not. It does that in your browser, and nothing is sent. If that split is wrong, you can ask Toast to read the document properly, and only then is it sent to our AI language-model providers. What is sent is the whole document — not just the part you marked as narration — because deciding which part that is is the thing being asked for. We do not store it: it is processed, cached for a week so re-reading the same file costs nothing, and not used for training. Loading a.txt, .md or .pdf from your computer still happens entirely in your browser; the file itself is never uploaded.
• Stock and generated media: When you add B-roll or background media, we send your text search query (or a text prompt for AI-generated images) to the corresponding media providers. We never send your footage.
• A story video from stock photos: If you make a story video from stock photos, our AI model writes a few search words for each scene from your script and Toast sends them to the stock-photo and video libraries to find a photo for it. The photos found, with the words describing the scene, are shown to our AI model to pick the best fit, and the chosen photos are kept in your library with their credits. Your script and your recording are never sent to the stock-photo and video libraries.
• How much of this a story video involves: Worth stating plainly, because the scale is different from adding a piece of B-roll by hand. A story video is illustrated end to end at roughly ten pictures per minute — about a hundred images for a ten-minute video — and each one is a separate request to the image provider carrying a description built from your script, including the narration line that picture illustrates. You approve the plan before any of it runs, and you can stop the run at any point.
• Generated images, stored on our servers: When you use Toast to generate images — for B-roll, or for a story video built from a voiceover — the resulting images are stored in our own cloud storage, together with the prompt that produced them, so you can reuse them instead of paying to generate the same picture twice. This is the one category of media Toast does keep. They are private to your account, served over expiring signed links, and deleted when you delete them. If the prompt was derived from your script, that line of your script is stored alongside the image and is sent to the image provider as part of the prompt.
• Pictures Toast drew are sent back to a model to be checked: Every picture Toast generates for a story video is checked against what you asked for — does it match the look you chose, does it show the moment it was meant to show, and is the character in it still the same character. That check is another AI model, so each generated picture is sent to it over an expiring link, together with the description that asked for it. This includes the reference picture Toast draws of each character, which is also read back to tell you what it actually drew. Each picture is sent more than once: a separate check looks for lettering, a signature or a printed border the image model added on its own (and, in a line-drawing style such as Whiteboard, a photograph pasted into the drawing), and a picture that fails it is drawn again and the new one checked too — so pictures you never see, because they were thrown away and redrawn, are sent to the model as well. A picture from your library that was made before this check existed is sent once before it is reused. A third check compares each picture with the facts its description states — the right things, where they are, who is not in it — and a picture that gets one wrong is drawn again from a rewritten description, up to twice, and the new picture is checked too. To write and rewrite that description, the scene description (built from a line of your script and the descriptions of your characters) is sent to the language model as well. Each kept picture is also looked at once for its layout: a model is shown the picture with a grid drawn on it and asked which parts hold people or the main subject, so captions and graphics are placed around them rather than on them; the answer is stored with the picture, so it is not asked again unless the picture is redrawn. These are pictures Toast made — never a photo you uploaded, and never a frame of your own video.
• Reference photos you upload: If you upload a photo to lock what a person, animal, object or place in your video should look like, that photo is stored on our servers and sent once to our vision model, which reads it and writes a description. That description — not your photo — is what gets sent to the image provider for every picture it draws. Your uploaded photo is never sent to an image generator, cannot be used as the basis for an AI edit, and is never offered back to you as stock imagery in another video. It stays as the picture we compare the generated frames against.
• Pictures you put on a frame yourself: If you replace a frame of a story video with a picture of your own, or use your own photo or logo for a person or company in it, that picture is stored on our servers with the rest of your images. It is not sent to any AI model or any other company, it is not checked or described, and it is never offered back to you as stock imagery in another video.
• Real people and companies your script or video names: When a story video's script names a real person or company, or the speaker in a talking-head video says one, Toast sends that name — just the name, never your script or transcript — to a free public knowledge base and its library of freely licensed media, to find a freely licensed photograph or logo, and, only when none exists, sends the company's web address to a logo service for its logo. For a talking-head video the name sent is the full name our AI model identified from your transcript, with a short description of who or what it is that we compare on our servers and do not send to the knowledge base. The picture found is stored with your images, placed into your video instead of being drawn by an AI, never sent to an AI model, and credited so you can list it in your video's description. When no freely licensed photograph of a person exists, that moment gets no picture rather than a stand-in.
• Uploading a photo of a person: Before you can upload a photo of a person, we ask you to confirm that it is you, or that you have that person's permission. We also check the photo, and we will not generate images from a photo of a child unless you have told us it is a photo of you. If we refuse it, the upload is deleted rather than kept.
• Editing a generated image sends that image to the provider: When you ask Toast to change something about a picture it made for you — “keep everything the same, just change her expression” — the only way to keep the rest of the picture is to give the provider the picture. So that request sends a temporary, expiring link to that stored image along with your instruction. It is always an image Toast generated for you: never a photo you uploaded, and never a frame of your video.
• Animated videos (when this feature is available to you): An animated video is made by an AI agent from what you type. To make one, your request, as you typed it, is sent to an AI model, directly to its provider, which may keep what it is sent for up to 30 days and does not train on it, with the format and length you chose, together with Toast's short descriptions of the looks your earlier animated videos used and the opening line each of them was narrated with, so a new one does not repeat them. If you ask to choose a direction first, the same request and descriptions are sent to the AI model to write three; all three are kept with your earlier looks whether or not you build one, and the one you choose is sent again when the video is made. Before you make a video, Toast writes its storyboard: the same request and descriptions, and the direction you chose (or three Toast writes to choose from, kept the same way), are sent to the AI model to write each scene's picture and words. Any words you change in it are sent to the AI model with the video, which narrates and shows them as you wrote them; the storyboard and your changes are kept with that video and deleted with it. If you ask Toast to write the storyboard again, or one scene of it, your note and the storyboard as you left it are sent to the AI model the same way, and every version is kept with that video and deleted with it. Toast also draws a picture of each scene on its own servers, from the page the agent wrote for it (nothing more is sent anywhere to draw it); the pictures are kept with the storyboard and deleted the same way. While the agent works it also sends the AI model the narration it writes and pictures of the video Toast is rendering for you, so it can judge them. The narration goes to our speech service to be spoken; for a language its fast voice does not speak, that spoken audio (made by the speech service, never a recording of anyone) goes back to it once to time its words. If the video has music, a few search words the agent wrote go to a music library, and the track's credit is shown with your video, as its licence requires. If you ask for a publishing kit (titles, descriptions and hashtags for the platforms you pick, and a designed thumbnail), the AI model is sent that video's request, its narration and look notes, its strongest frames and a sheet of its frames (pictures of the video Toast rendered), the platforms and sizes you picked and, if you choose "Other", the platform name and notes you typed. Toast adds the music credit to each text itself; the kit is kept with your video. If you ask for the same animated video in another shape (tall for short-video apps, or square for a feed), the AI model is sent that video's request, the project Toast's agent wrote for it (its code, with the narration text in it, and its look and notes) and pictures of the new version as it is rendered. Its narration audio and music stay on Toast's servers: nothing goes to the speech service or the music library. If you change scenes of an animated video (you stop it at a frame and say what to change, for one scene or several at once), the AI model is sent what you typed, the project Toast's agent wrote for that video (its code, with the narration text in it, and its look and notes), each frame you stopped at, any pictures you add for the change (your own material: the same upload and check as for a new video, kept and deleted the same way) and pictures of the changed video as it is rendered. If the change is to what is said, the new line is sent to the speech service to be spoken and heard back once; nothing goes to the music library. If the change is only to a spoken line, no AI model is sent anything for it; the new line goes to the speech service as above. Only those scenes' seconds are replaced in your video. What you typed and each changed version are kept with the video. If you ask for a hero moment in a talking-head or story video (one short animated graphic laid over a few seconds of it), the AI model is sent the words said in those seconds, how long they last, the video's shape, whether a person or a picture is on screen, and the video's look described in words. No frame of your video is sent, and nothing goes to the speech service or the music library. The clip Toast makes, and what was asked for it, are kept until you delete the project or your account. A hero moment in a property video sends the same AI model, directly, the moment's words (the line you type, or the facts you pick for it as you typed them, such as the bedrooms and locality, or the price and the words over it), its length, the video's shape and its look in Toast's words. Never a picture, a still, a frame or your footage. What was asked for it and its clip are kept until you remove it, or delete the video, the property or your account. A premium highlight (at most one a video: a hero moment for one highlighted item) sends the same AI model, directly, in the same way, the item's name as it is on its tag (Toast's word for it, or the one you typed when you tapped it) and how long its shot lasts. Never a picture, a still, where the item is in the shot, or your footage.Audio you speak instead of typing. When you give Toast a recording of what you want, Toast sends it to a speech-to-text service once to write down its words and shows them to you to correct. Nothing of the recording is kept: only the request you then send, as with typed words.A motion video made to your own audio. When you choose to make a video to a recording (a voiceover, a podcast clip, a reel's sound), you first tick that it is your own audio or that you have permission to use it. Every recording you give is first sent to a speech-to-text service once, and not kept, so Toast can suggest which of the two you mean. Toast stores the recording privately and sends it to the speech-to-text service to write down its words and when each is said; those words and times go to the AI model directly, which designs the video to them. The recording itself becomes the video's soundtrack, with only its loudness changed. It is never used to copy anyone's voice. It is deleted with the video and with your account, and a recording no video uses is deleted within a day. When you pick a video file, your device takes out its sound: the video itself is never uploaded.Screen recordings you make an animated video from. To understand them, Toast sends each recording's sound (if it has any) to a speech-to-text service for the words and when each is said, and pictures of each recording, up to one every four seconds and at each change of screen, to an AI model, with the words and what Toast read from the video's other recordings, to say what each screen is and the steps they show. The AI model is reached through a service that asks it not to keep or learn from what it is sent. Each recording is read once: adding one to a video later sends only that one. Your screen may show your own or other people's details, such as emails or names: these pictures carry whatever was on screen. If your recording shows your face, such as a face bubble, these pictures carry it too; Toast's own servers mark where faces are in them (a box, never who it is), only to keep words and graphics off them, and these boxes are stored and deleted with the recording and sent to no one. To make the video, Toast's motion agent, on Toast's own servers, receives the recordings (or the parts of them it chooses) and draws motion graphics over them, such as arrows, boxes and zooms. It sends an AI model what Toast read on each screen, the words said in the recordings, still pictures taken from them (to choose what to show and where to point) and pictures of the video it makes (to check it), but never a recording itself as video. The AI model's provider may keep these for up to 30 days. Screens you mark ‘Leave it out’ are removed, and screens you mark ‘Blur it’ are blurred, on Toast's servers before the agent receives the recordings. If you ask for an animated video about a website or a pull request, Toast's server first reads it for you. For a web page, that is the page itself and, for a product's or business's home page, up to five more of its pages (such as pricing, features or a menu), fetched the way an ordinary visitor who is not logged in would, and only where the site's robots.txt allows; for a public pull request on a code-hosting site, its description, commit messages, list of changed files, the issues it says it closes and its changes to the code. The link you give goes to that site or to the code-hosting site, and nothing else of yours goes with it. When a page shows little until its scripts run, or when you ask for screenshots of it, Toast's own server also opens it in a browser, the way a visitor who is not logged in would, without clicking or typing anything, to read the text it then shows and, if you asked, to take screenshots of it. Only the address you gave goes to the site, as when Toast reads it. You see every screenshot before you make the video and can remove any of them. Those you keep are treated like pictures you upload for an animated video: each is checked once by an AI model for people in it, stored privately with your video, sent with your other pictures to the AI model that makes it, and deleted with the video. Screenshots of a video you never make are deleted within a day. If you ask Toast to use your logo or pictures it found on the pages it read, Toast's server fetches them from the site the way a visitor who is not logged in would; you see each before you make the video, and those you keep are treated like pictures you upload for an animated video. Reading it to show you what was found keeps nothing. To tell you what kind of page it is (a product, an article, a paper, an event...) and which kinds of video fit it, the page's address, title and description and the start of its text are sent to an AI model, through our AI routing service, which asks it not to keep or learn from what it is sent; this happens each time a page is read, and only the kind of video you chose and a one-line description of the page are kept with your video. When you start the video, what was read (the pages' text, or the pull request's text, file list, linked issues and a few pieces of its code that Toast chose) is sent to the AI model that makes animated videos with your request and kept with your video. Before that, Toast removes anything that looks like a password or key from the text and leaves out any piece of code that contains one. If you choose to make the video in the colours Toast found on the site, those colours are sent too. If you make an animated video from text you give (pasted, or a PDF or text file), a file is read in your browser and never uploaded; only its text is used. When you start the video, the first 40,000 characters of that text are sent to the AI model that makes animated videos with your request, after Toast removes anything that looks like a password or key, and are kept with your video until you delete it or your account. That text does not go through our AI routing service or anywhere else. If you make a series from a long document (one animated video per chapter), the document is read and split into chapters in your browser and is never uploaded. Toast keeps only the list of chapters (their titles, page numbers and sizes). A chapter's text (up to 40,000 characters) is sent to the AI model only when you make that chapter's episode, in the same way as text you give, and is kept with that episode until you delete it, the series or your account. Chapters you never make are never sent, so a whole book is sent only if you make every chapter. If you add your own material to an animated video (pictures such as a logo, a product shot, a screenshot or a chart, and lines of text you want on screen word for word), each picture is sent to the AI model once when you add it, so Toast can tell you what it shows and whether it shows people. A picture of people needs you to tell us it is you or that you have their permission, and a picture of a child is refused unless it is of you; a refused picture is not kept. Otherwise it is kept in your private library, with its camera and location details removed, and deleted with the video that used it (unless another of your videos still uses it) or with your account. When you start the video, your pictures, your notes about them and your lines of text are sent to the AI model with your request, and are kept with your video. None of it goes through our AI routing service or to an image model. A picture you add as a style reference (a frame or a poster whose look you like) is checked, sent, kept and deleted the same way; the AI takes only its look from it, and it is never shown in the video. If you make an animated video in your brand, your brand kit's business name, website, colours and fonts and its logo are sent to the AI model with your request in the same way, so every scene uses them; the first time, the logo is also sent once so Toast can check whether it shows people (a logo of people is refused there: add it as your own material instead). The kit's words are kept with your video; the logo stays in your brand kit and is not deleted with the video. Nothing of your brand kit is sent for a video you make without it. No video of yours is ever part of an animated video, and nothing else you upload is, apart from your brand kit's logo when you make a video in your brand.
• Account data: Email address and authentication tokens (held by our sign-in provider) for account management.
• Payment data: When you buy credits you are sent to our payment provider's own hosted checkout page and enter your payment details there. Toast never receives, sees or stores your card, UPI or bank details. What we keep is the payment identifier, the amount and the time, which is what lets us credit your account, show you a receipt and refuse to charge you twice for the same purchase. When you add money to a wallet the same is true, and we also keep your wallet's currency and a record of every amount added and spent, by which video, so your balance and history are right. To choose the currency a new wallet opens in, Toast's server reads the country your request comes from: it is looked up from your IP address in a table on Toast's server, and neither is stored. Nothing is sent to anyone for it. (IP to country data by DB-IP, under CC BY 4.0.)
• Project metadata: Project names, edit manifests, and version history are stored on our servers. These contain text and timestamps — never video or audio content.
• Your transition choices: When you export, Toast records which transitions you kept, changed or removed from the ones its AI suggested, and which you added yourself — only the names of the transition types, never your words or timings. Toast uses this on its own servers to suggest the transitions you prefer; it is not sent to any AI provider, and it is deleted with the project or your account.
• Feedback attachments: If you attach a file to a feedback or suggestion form, it is emailed to our team via our email provider. We accept images and text files only; video attachments are rejected. If you send feedback from the Feedback button, the address of the page you were on (for example toast.video/editor/ followed by your project's ID) is included in the email, unless you untick it.
• Video ratings: If you rate videos on our public rating page (toast.video/rate), we keep your answers: the rating you chose, the reasons you tapped, the moment you marked, any short notes you wrote, and the time. They are tied only to a random code your browser makes for the page, never to your name, email address, account or IP address, unless you choose to link your contact details to them (below), and we remove email addresses and phone numbers from your notes before storing them. Nothing is sent to any AI provider or other third party; we use the answers to measure and improve the videos Toast makes.
• Contact details on the rating page: If you choose to leave your name, email address or phone number after rating videos on toast.video/rate, Toast keeps them apart from your answers, unless you tick to link them, so we can contact you about your feedback. They are used for nothing else, are not sent to or shared with anyone, and are deleted after 12 months, or sooner when you remove them on the same page in the same browser or ask us to.
• Analytics: We collect product-usage analytics (page views, feature usage) through our analytics service. Events are tied to a pseudonymous account identifier (your sign-in account ID) so we can understand feature adoption; we do not add your name or email address to that identifier. We record how pages are used so we can find what goes wrong and make Toast better. These recordings show what is on the page and what you type, except passwords: your name and email address, scripts, transcripts, project names, and pictures such as project covers, which can be taken from your video. They also hold the page's log messages, and analytics events can carry the text of what you clicked. Video playing on the page is not recorded. This applies wherever this page says something is sent to no one or to no third party: what appears on a page can be in a recording, tied to your account ID.

The rest of this page describes each service by what it does (an AI model, a speech-to-text service, our storage provider), because the company doing it may change. This section is the one place we name them: the companies in each role today, each with its role in italics. Each receives only the data described above for the stated purpose. None receives the video you edit; a property shoot you upload is stored on Cloudflare R2 (below), and when you ask Toast to read it, stills of it go to Anthropic through OpenRouter and its sound to ElevenLabs (section 3); its files themselves go to no one. The footage of an event you upload is stored on Cloudflare R2 too, and when you ask Toast to read it, stills of it and what its sound holds go to Anthropic through OpenRouter and its sound to ElevenLabs (section 3); its files themselves go to no one.

AI editing

• Anthropic (Claude), through OpenRouter (an AI model, through our AI routing service), run by Anthropic or, when Anthropic's own service is unavailable, by Amazon Web Services (Bedrock) or Google Cloud (Vertex AI), which run the same Claude models — AI editing decisions (transcript text + prompts). Toast also calls Anthropic directly, but only as a fallback when OpenRouter is unavailable: the stock-footage search words and image prompts Toast writes from your transcript
• Anthropic (Claude), directly, for animated videos (the AI model that makes animated videos) — your request, the looks of your earlier animated videos, the narration the agent writes, and pictures of the video being rendered for you (never your own video); for your own material, the pictures you add (each also checked once when you add it) and the lines of text you type; for a publishing kit, frames of that video and any notes you type about another platform; for the same video in another shape, the project Toast wrote for it; for a change to one or more scenes, what you typed, that project, each frame you stopped at and any pictures you add for it; for a hero moment in a talking-head or story video, the words said in it, its length and the video's look in words (never a frame); for a hero moment in a property video, its words (a line you type, facts you pick, as typed, or a highlighted item's name), its length and the look (never a picture or your footage); for a video made to your own audio, the words and timings of that audio (never the recording itself); for a video made from your screen recordings, what Toast read on each screen, the words said in them and still pictures taken from them (never a recording itself as video); for a video about a website or a pull request, what Toast read from it; for a video from text you give, that text (up to 40,000 characters); for a series, the text of each chapter you make an episode of. Anthropic keeps API inputs and outputs for up to 30 days and does not train on them
• GitHub (the code-hosting site) — when you ask for an animated video about a pull request, Toast asks GitHub for it (the repository and number in your link); nothing else of yours
• OpenAI (an AI model) — AI editing fallback (transcript text + prompts)
• OpenRouter (our AI routing service) — AI model routing for the above, configured to deny provider data retention/training (transcript text + prompts). Also routes the vision model that checks generated pictures and reads uploaded reference photos, so images reach it too, not text alone; for an animated video about a link, the start of that page's text, to say what kind of page it is; and, when you ask Toast to read a property, stills of your property shoot to Anthropic's models; when you ask for a property video, the sentences said to the camera in that shoot; when Toast makes it, stills of the finished video with its room names and the sentences said over them, to check it, and pictures of the finished video with its words and cards on it, to judge whether it reads; and, when you ask Toast to read an event, stills of your event footage with what its sound holds (who speaks when, music, applause, laughter) to Anthropic's models; and when you ask for an event video, the sentences said in its clips, never the names you typed; and, when you make an animated video from your screen recordings, pictures of those recordings with the words said in them, to Anthropic's models

Audio

• ElevenLabs (our speech service, and the speech-to-text service) — Transcription (including the sound track of each file of a property shoot or an event you ask Toast to read), audio noise removal, and — if you type a script instead of uploading a voiceover — speaking that script aloud. In that case the words you wrote are sent to ElevenLabs, which is the same kind of content as your transcript. For an animated video, it speaks the narration the agent wrote from your request (and, when you change what a scene says, its new line), and it hears every spoken line back once (its own stock voice, never a recording of yours) to check the words were said as written and, for a language its fast voice does not speak, to time them. If you tell Toast how one of your names is said (for example “Hatke” said as “Hut-kay”), we keep that spelling with your account and send it to ElevenLabs in place of the name whenever an animated video of yours speaks it; your captions and the screen keep the name as you write it. If you press play beside a name, that name, or how you typed it should be said, is sent to our speech service to be spoken in the narrator's voice; nothing of it is kept. Audio you give Toast for an animated video (its words): a spoken request once, and a soundtrack once more with when each word is said. The sound of each screen recording you make an animated video from, once.
• ElevenLabs, your own voice (our speech service) — if you choose to narrate a story video in your own voice, a recording of your voice is sent to ElevenLabs each time a narration is spoken in it, to make a temporary copy of the voice that is deleted as soon as that narration is spoken. Toast checks the recording once with ElevenLabs' transcription to make sure only one person is speaking. Toast does not store the recording: it stays on your device, and our server keeps only a record of your consent (which promise you ticked, when, and a fingerprint of the recording). “Delete my voice” ends it at once; no copy of your voice is kept at ElevenLabs to delete.
• CleanVoice (the noise-removal service) — Audio noise removal (audio only)

Media

• Replicate (our image provider, running Black Forest Labs' Flux models, and ByteDance's Seedream model when you choose “Best” picture quality for a story video) — AI image generation and editing. Text prompt for a new image; for a story video that prompt is derived from your script, so it can contain lines of your narration. On a “Standard” video, a picture our own check finds wrong is redrawn once with Seedream too, so its prompt goes there as well. Editing an existing image also sends that image (an expiring link to a picture Toast generated for you) — never your video, never an upload
• Pexels, Pixabay (stock-photo and video libraries) — Stock footage and images. Search words only: what you type into a stock search box, and, for the B-roll in a talking-head video, a few words the AI picked from what you said (such as “city skyline at night”), sometimes reworded by the AI before the search, and, for a story video made from stock photos, the search words for each scene. Never your video, your audio or your transcript
• Jamendo (the music library) — Background music search and download. Usually a few words Toast worked out from your video. But if your script document has a music cue sheet in it, what we search for is the cue you wrote — the keywords out of your own sheet, sent without you pressing search. A few words of your document, never the document, and only when you put a cue sheet in it. For an animated video, a few search words the agent wrote from your request; only Creative Commons Attribution tracks are used, and their credit is shown with your video. For a property video, a mood Toast's planner picked from a fixed list (such as “warm” or “elegant”), sent as Toast's own search words, never your words or your footage; the track is fetched by Toast's servers and its credit is kept with the video. An event video's music is chosen the same way

Real people and companies

• Wikidata and Wikimedia Commons (the free public knowledge base and its library of freely licensed media) — the name of a real person or company your script or video names, to find a freely licensed photograph or logo; just the name
• Brandfetch (the logo service) — a company's web address, only when no freely licensed logo exists

Platform

• Clerk (our sign-in provider) — Authentication (Google OAuth, magic links). If an account breaks our terms, we can block it. Our sign-in provider then records that the account is blocked, and we can undo this.
• PostHog (our analytics service) — Product analytics and page recordings, which show what is on the page and what you type except passwords, including your name, email address, scripts and pictures such as project covers (tied to your account ID)
• Sentry (our error tracking) — Error tracking (with PII redaction), and a daily alert to us when an account's uploads pass a size or count limit (the account id, the gigabytes and the number of files; never what was uploaded)
• Razorpay (our payment provider) — Payment processing for credit purchases and wallet top-ups. You are sent to Razorpay's own hosted checkout page and enter your payment details there; Toast never sees or stores your card, UPI or bank details. We keep the payment identifier and the amount so we can credit your account and show you a receipt
• Resend (our email provider) — Transactional and feedback email delivery
• Vercel (a hosting provider) — Frontend hosting
• Railway (a hosting provider) — API hosting; hosting for Toast's own service that prepares the property and event files you upload and makes your finished property videos from them (it keeps nothing once its work is stored); and hosting for the service that draws graphics cards. That service receives the text that goes on a card (and, for a card that shows one, a picture Toast already keeps for you) and returns a clip; it never receives your video, and it is ours, not a third party that gets to keep anything
• Supabase (our database provider) — Database (project metadata only)
• Cloudflare R2 (our storage provider) — Storage for images Toast generated for you, for graphics cards and caption clips it drew for you, for reference photos you upload, and for pictures you add to an animated video as your own material, and for the videos, photos and audio of an event you upload and the videos, photos and audio of a property shoot you upload, and the finished property videos Toast made from them. Never the video you edit.

Project metadata is retained as long as your account is active. You can delete any project at any time, which removes all associated metadata from our servers. Browser-local data (OPFS files) is managed entirely by your browser.

A reference photo you upload is kept until you delete it or delete the subject it describes, for the same reason: it is what later videos compare their frames against. Deleting it removes it from your library immediately.

A picture you add to an animated video as your own material is kept until you delete the video that used it (unless another of your videos still uses it) or your account. A picture you add but never make a video with is kept until you delete your account.

Generated images are kept until you delete them, and deliberately outlive the project they were first made for — that is what lets a later video reuse them rather than regenerate them. Deleting one removes it from your library immediately, and the file itself is erased from our storage within 30 days. We keep a record that the image existed after that, with no picture attached, because an image you later edited from it has to remain explicable. We do not control how long the image provider retains a prompt or its output on their side; see the image provider's own policy for that (section 4 names it).

An animated video Toast made for you, its preview pictures, its notes, its publishing kit, its other shapes, its changed versions (with what you typed for each), a hero moment's clip and what was asked for it, and the project it was made from (its code, its narration audio, its music and their timings, kept so the video can be made in another shape or changed) are kept on our storage until you delete the video or your account; deleting the video removes all of them and the notes it added to your earlier looks. What its agent sent the AI model is kept by the model's provider for up to 30 days. Toast's owner can see each animated video, its request and what it cost to make, to review the quality and the cost of the animated videos; nobody else at Toast, and nothing about it is sent anywhere.

The videos, photos and audio of a property shoot, and the copies Toast made from them, are kept while the property is in your account. Deleting the property, or one of its files, deletes them (a file you used in two of your properties is kept until both no longer use it). The finished videos Toast made of a property are kept while the property is in your account and deleted with it.

The videos, photos and audio of an event, the copies Toast made from them and the words heard in their sound, are kept while an event in your account uses them. Deleting the event, or one of its files, deletes them (a clip in two of your events is kept until neither uses it). What Toast understood when it read an event is kept with the event and deleted with it.

Deleting your account erases the files too. Everything we have stored for you — every generated image, every reference photo and every file of a property shoot or an event — is removed from our storage along with your records, whether you delete the account yourself or it is removed on the sign-in provider's side.

When you delete your account, we keep a scrambled code made from your verified email address. It cannot be turned back into your address, and we use it only to stop the same address getting welcome money twice. We delete it after 2 years.

If you join through a friend's link, we keep a record of who invited you so we can give you both your reward. It is deleted when either account is deleted.

A story video contains no camera footage at all. The rule above is not weakened by any of this — for that kind of project it is not even engaged.

You may request deletion of your account and all associated data, apart from the scrambled code described in section 5, by contacting us at gopikrishna@toast.video.

We may update this policy from time to time. Material changes will be communicated via email or an in-app notice.

Arnanda Technologies Private Limited. Questions? Reach us at gopikrishna@toast.video.