What’s Changed: Sakura AI is still online, but its privacy policy is still dated January 6, 2024, and it never mentions chats. Its terms require users to be over 18, yet Google Play rates the app 12+. It is safe enough for adults who keep personal details out of their chats, and a poor fit for teenagers.
Is Sakura AI safe? For adults who keep their real identity out of the chat window, mostly yes. Its paperwork gives you far less to go on.
Sakura’s privacy policy, last updated on January 6, 2024, never mentions chats, conversations or messages on a site built around them.
Its App Store listing says “Other User Content” may be linked to your identity and used for “Developer’s Advertising or Marketing”.
None of that proves Sakura misuses what you write. I still keep my real name and town out of every Sakura chat, since no written rule stops it from doing so.
This guide covers what Sakura can see, which Reddit allegations hold up, what the free tier asks for in return and whether a paid plan earns its price.

Is Sakura AI Safe to Use?
Sakura AI is reasonably safe for adults who assume the company can read every chat. A named Florida company runs it and no breach has been reported, but its policies are too thin to trust with anything private.
Sakura AI, LLC gives its address as 1901 Starboard Way in Jacksonville, Florida, in its terms of service. Its Google Play listing adds a second Florida address in Tarpon Springs and a phone number.
That is more accountability than plenty of anonymous chat sites offer.
The terms and the privacy policy both carry the line “Last updated: January 06, 2024”, the latest revision either one shows. I treat a policy that old on a chat app as a warning sign.
If you would rather pay for a companion than earn messages through Discord tasks, Candy AI is the one I point most people to, at $13.99 a month or $47.88 for a year.
Its free tier gives you 5 messages to judge it, and our Candy AI safety check covers its policies.
Can Sakura AI See Your Chats?
Assume Sakura AI can read your chats. Your chat history is stored with your account, and the privacy policy sets no limits on access to it because it never mentions chats at all.

The privacy policy covers cookies, usage data, California privacy rights and five social logins: Google, Facebook, Instagram, Twitter and LinkedIn.
It names PostHog and Google Analytics as its analytics services, but it never uses the words chat, conversation, message, moderation or encryption. Discord is missing too, even though Sage, Sakura’s route to extra free messages, runs through it.
Its California privacy section even marks sensitive personal information as collected, giving “Account login and password information, geolocation data” as the examples. I read that line as an unedited template, which leaves the real handling of your chats undescribed.
The app stores fill in part of the gap. Apple’s App Store listing says your email address, “Other User Content” and user ID may be linked to you and used for “Developer’s Advertising or Marketing”. On an app where nearly everything you make is text, I take that category to cover your chats and characters.
Google Play’s data safety page says email, device IDs and diagnostics are shared for analytics. It also says data is encrypted in transit and that you can ask for it to be deleted.
Is the Sakura AI Discord Login Safe?
Signing in with Discord is standard and does not hand Sakura your Discord password. The bigger exposure is the Sakura Discord server, which you have to join to earn Sage, the status that unlocks extra free messages.
Sakura’s website now loads its sign-in code from Clerk, a third-party login service, at clerk.sakura.fm.
In a December 2025 thread, a reply tells the worried poster that Sakura uses “discords login via auth0”, which no longer matches the site’s code, though its point about passwords stands.
Discord stops being optional once you want Sage. Sakura’s upgrade page tells free users who “Keep reaching your daily message limit” to “Apply in Discord” for Sage. A separate prompt on the site asks you, “To continue chatting,” to upload a screenshot “showing that you’ve watched and liked the video”.
I would join that server with a spare Discord account that carries no real name, since strangers in the server are a bigger risk than the login.
Was Sakura AI Hacked or Caught Selling Data?
No breach of Sakura’s own systems has been confirmed, and Have I Been Pwned, the public breach database, lists none. A June 2025 message sent across Sakura’s Discord server claimed it sells user data, and no breach record or company document supports that.
A June 2025 thread relayed a message that a supposed ex-employee sent to the entire Sakura Discord server.
It claimed Sakura was “selling our data” and urged people to delete their accounts and change passwords. A reply in the same thread said the server “just got hacked right now”.
I don’t buy the data-selling claim on that evidence. The relayed message names no buyer, it landed while the server was reportedly being hacked, and Have I Been Pwned lists no Sakura breach.
Sakura’s privacy policy says “We do not sell personal information as the term sell is commonly understood”. My read is that this wording leaves room for sharing that a broader legal definition of selling would cover.
Is Sakura AI Safe for Teenagers?
No, Sakura AI is not safe for teenagers. Its terms say users must be over 18 and its settings include a “Show NSFW (18+)” switch. Its store ratings and privacy policy point to younger users, which makes that rule easy to miss.

Sakura’s terms, privacy policy and store listings give four different answers on age, and only the strictest one fits an app with an adult content switch.
| Where | What it says about age |
|---|---|
| Terms of service | Users must be “over the age of 18” |
| Privacy policy | An opt-in rule for users “between 13 and 16 years of age” |
| App Store | Rated 16+ |
| Google Play | “Rated for 12+” with “Parental Guidance Recommended” |
Google Play’s 12+ rating contradicts Sakura’s own terms, and it is the rating a parent browsing Google Play sees. My answer for a parent asking about this app is a flat no.
Regulators are watching this category. On September 11, 2025, the FTC ordered seven companies that run consumer AI chatbots to report how they measure and monitor potential negative effects on children and teens.
Sakura was not among the seven, but its documents would struggle with the questions the FTC asked. Nothing in its privacy policy explains how it checks age or keeps minors away from the NSFW setting.
Does Sakura AI Have a Filter or Ban Accounts?
Sakura AI lets adults switch NSFW content on, but it is not filter-free. Its terms ban “obscene” content and let Sakura suspend any account “for any reason whatsoever”, so private roleplay runs on the company’s tolerance.
The terms forbid content that is “unlawful, offensive, upsetting, intended to disgust, threatening, libelous, defamatory, obscene or otherwise objectionable”.
The same terms let Sakura “terminate or suspend Your Account immediately, without prior notice or liability, for any reason whatsoever”.
Reports of hidden limits go back years. In a December 2023 thread, a reply says Sakura has a “secret filter” for banned topics and guesses that tripping it too often gets an account locked.
A February 2024 thread accused the Sakura Discord moderators of threatening members who questioned a moderation call. One reply in the same thread said the moderators were not part of the dev team, and another said they “all got kicked out”.
That termination clause is why I keep a copy of any character I care about outside Sakura. The terms also say “the Company does not guarantee there will be no loss or corruption of data”.
If a hidden filter is your main worry, these no-filter alternatives put fewer limits on roleplay.
Is Sakura AI Free and What Do the Paid Plans Cost?
Sakura AI is free with a daily message limit. Sage adds free messages in exchange for social media tasks, and the web plans cost $19 a month for Diamond and $39 a month for Infinite.
Sage is less generous than its reputation. The upgrade page pitches it as “Free Messages” for people who keep hitting the daily limit, and it never promises unlimited chat.
In a thread on Sakura’s restrictions, one person who earned Sage said it was “still giving me a message limit”.
| Plan | Web price | iPhone app price | What it adds |
|---|---|---|---|
| Free with Sage | $0 | $0 | A daily message limit, with extra free messages for social media tasks |
| Diamond | $19 a month | $19.00, plus a $131.99 option | Unlimited messages, the Dragonfruit model and Fusion Mode, which offers several replies per message |
| Infinite | $39 a month, marked as a sale | $39.00, plus a $274.99 option | The Glowing Dragonfruit model, “Infinite Memory” and everything in Diamond |
The larger App Store prices line up with the 40% yearly saving the web page advertises, so they are most likely annual plans. The iPhone app also sells a $5.00 Gold tier that the web page does not show.
Plans bought from Sakura directly are “non-refundable” except where the law requires otherwise, and the terms leave other refunds to “a case-by-case basis”. App store purchases follow Apple’s or Google’s own billing terms instead.
Diamond is the plan worth a month of my money, since unlimited messages fix the main free-tier complaint. Infinite costs double, and I would want proof of its memory claim before paying that.
Why Do Sakura AI Chats Feel Repetitive or Forgetful?
Sakura AI chats feel forgetful and repetitive most likely because its models keep only a short stretch of the conversation in view. The fix Sakura sells is the Dragonfruit model on Diamond, which promises “better memory”.
The same complaint shows up in threads from 2025 and 2026. A reply in a July 2025 thread says the bots “remember only like 8 messages”, and a March 2026 post lists “lots of repetition, lack of immersion, and loss of memory”.
A June 2026 post says Sakura “feels soulless” and “doesn’t rly have good memory”. I restate the scene in one line every ten or so messages on the free tier, which keeps a story on track for longer.
The apps look stalled as well. The iPhone app is still on version 1.1.1 from May 23, 2024, and the Android app was last updated on December 29, 2024.
| Symptom | Likely cause | Fix |
|---|---|---|
| The character forgets earlier scenes | Short memory on the free models | Restate key facts every 10 or so messages |
| Replies repeat the same phrases | Long chats drift into loops | Regenerate the reply or start a fresh chat |
| A daily message limit pop-up | Free allowance used up | Wait for the reset, apply for Sage or try Diamond |
| “Failed to delete character” | Other people are using your bot | Edit the character instead of deleting it |
If memory matters more than the character library, Janitor AI vs Chub AI weighs two other big character sites on library size, content rules and setup hassle.
How Do You Use Sakura AI Safely?
You use Sakura AI safely by keeping identifying details out of chats and characters, joining its Discord with a spare account, and saving copies of anything you would hate to lose.
This is the routine I use on any chat site whose policy says nothing about chats:
- Keep your full name, workplace or school, town and photos out of chats and character bios.
- Join the Sakura Discord with a spare account, and turn off direct messages from server members in Discord’s privacy settings.
- Turn on two-factor authentication for that Discord account and for the email you use on Sakura.
- Copy any character definition or long story you care about into a local file.
- Ignore server-wide messages telling you to delete your account or join another server until Sakura confirms them in its official channels.
- Buy through the App Store or Google Play if you want their refund process, since Sakura’s own terms make direct purchases non-refundable.
Persona text is the easiest place to overshare. Here is the same persona written two ways.
Before: “I’m Jess Carter, 24, a nurse at St. Luke’s in Boise, and I live alone near Ann Morrison Park.”
After: “I’m Jess, a night-shift nurse in a quiet mountain city who lives alone and loves late walks.”
The second version gives a character everything it needs for a story and gives nobody a way to find you.
Is Sakura AI Worth It?
Sakura AI is worth it as a free character playground for adults who keep personal details out of it. It is not worth a yearly plan while its policies date from January 2024 and its apps have sat untouched since 2024.
I’ll give Sakura this much: the free tier is usable for casual roleplay, and the company behind it is easy to identify. Use it if you want a big library of user-made characters and do not mind earning messages through Discord.
Skip it if you are under 18, if long memory matters to you, or if you need a written promise about what happens to your chats. The best Character AI alternatives are where I would start looking instead.
If memory is what you want most, Nectar AI is where I send long-story roleplayers: its $34.99 Ultimate plan adds long-term memory and a 64K context window. Its free plan gives you 15 messages a day to judge the writing first.
Quick Takeaways
- Sakura’s privacy policy, dated January 6, 2024, never mentions chats, so treat everything you type as readable by the company.
- Its terms say over 18 while Google Play rates it 12+, and its NSFW switch makes it a poor fit for teenagers.
- The data-selling claim rests on a June 2025 Discord message, and Have I Been Pwned lists no Sakura breach.
- Free chat runs on a daily limit plus Sage’s social media tasks, and Diamond at $19 a month is the plan to test before Infinite.
- Keep identifying details out of personas and copy any character you value, since the terms do not guarantee your data against loss.

The point about the privacy policy never using the words ‘chat,’ ‘conversation,’ or ‘message’ is striking, especially since the California section lists sensitive data like login and geolocation. That gap suggests the policy may be a generic template rather than a description of actual data flows. I wonder whether the app store’s ‘Other User Content’ label is the only place chat data is even implicitly acknowledged, and whether that alone is enough for users to make informed choices.