Is Bluesky Safe? Privacy, Kids, Scams and What Moderation Catches
Bluesky is safe enough for most adults if you know one thing going in: almost everything you do there is public. Your posts, likes, follows and blocks can be seen by anyone, DMs are not end-to-end encrypted, and there's no private account.
On the plus side, porn and graphic content is hidden by default, you can subscribe to shared block lists, and moderation moves fast on spam. I sampled 25,000 random posts made in one hour on November 15, 2026, and checked each one about 25 hours later. 2.8% were already gone because the account behind them had been taken down, and 1.9% of the rest carried an adult-content label.
Below: what's public, the age rules, what my sample showed, and how to avoid getting phished.
(Disclosure: I make Agent Sky, a Bluesky growth tool. It has nothing to do with Bluesky's moderation. More on it at the end.)
How Safe Is Bluesky, Risk by Risk?
Each risk below comes with what Bluesky does about it and what you can do.
| Risk | How Bluesky handles it | What you should do |
|---|---|---|
| Privacy | Posts, profile, likes, following and blocks are public by design. No private accounts. | Post as if a stranger is reading. |
| DMs | Stored unencrypted; Bluesky can read them "for Trust & Safety purposes". By default only people you follow can message you. | Don't send anything you'd hate to see leaked. |
| Adult content | Porn, sexual and graphic content hidden by default. Can't be turned on with a declared age under 18. | Leave it off for teens' accounts. |
| Kids | Minimum age 13 plus local law. Age checks in the UK, Australia and several US states. No parental controls. | Better for older teens than young kids. |
| Scams and phishing | Fake "support" DMs and phishing links; 4,500+ accounts compromised in October 2026. | Turn on 2FA. Bluesky doesn't contact you about your account by DM. |
| Harassment | Reports, blocks, muted words, shared block lists, reply controls. | Subscribe to a block list and set who can reply to you. |
| AI training | Bluesky says it doesn't train generative AI on your posts. Others can scrape public posts. | Assume anything public can be collected. |
Sources: Bluesky's privacy policy and terms of service (both updated August 14, 2025), its age assurance post, and the Bluesky app's open-source code.
Is Bluesky legit? Yes. It's a real company that started as a Twitter project in 2019 and became independent in 2021, its app and protocol are open source, and its 2025 transparency report counts 41.41 million users at the end of 2025. Most of the "is it safe" worry is about the points in the table, not about whether the app itself is shady.
What Can Strangers See About You on Bluesky?
Nearly everything except your DMs, your mutes and your email. Bluesky's privacy policy says it plainly: "most user activity is public by design. This includes your posts, profile, likes, following, and blocks." Bluesky runs on an open protocol where every account's data sits in a public repository, so any app or script can read it, including the ones that show who blocked whom. (Mutes stay private, but mute lists are public.)
One setting helps a little, but it doesn't make you private: "Ask apps and sites not to show my account to logged-out users" in Settings > Privacy and security. The app is honest about the limit: "Other apps may not honor this request. It doesn't make your account private." In my sample, 22.1% of posts came from accounts with it turned on.
DMs are the other big one. The privacy policy says DMs "are unencrypted and can be accessed for Trust & Safety purposes." I go into the details in are Bluesky DMs private, and the workarounds for wanting a closed account are in can you make a Bluesky account private.
Is Bluesky Safe for Kids and Teens?
It's built for adults and older teens, not young kids. Bluesky's terms say: "To use Bluesky, you must be at least 13 years old and meet the minimum age required by your local laws." Some places set a higher bar, per Bluesky's own age assurance post:
- United Kingdom: age checks since July 2025 to keep children away from adult content.
- Australia: under-16s blocked since December 10, 2025. 16- and 17-year-olds can't see adult content or use DMs.
- Virginia: under-16s blocked from January 1, 2026. Tennessee and Mississippi: under-18s blocked.
- Brazil and Texas: some content and features restricted for under-18s (2026 updates).
Bluesky's 2025 transparency report counts 364,960 age-assured accounts by the end of 2025.
The protections that apply everywhere come from the app's defaults. I read Bluesky's open-source app code to check them:
- Adult content is off for every new account. Posts labeled porn, sexual or graphic media stay hidden until you switch it on in Settings > Moderation tools. (Posts labeled nudity are a separate setting and are shown by default.)
- If your declared age is under 18, you can't switch it on. In regions with age checks, accounts without full access can't turn it on or use chat either.
- On iPhone, the switch is greyed out. The app says adult content "can only be enabled via the Web at bsky.app".
- DMs default to people you follow. Strangers can't message a new account unless that setting is changed.
- Logged-out visitors see no posts with adult or nudity labels.
What the defaults don't fix: Bluesky has no parental controls or supervised accounts, so a teen can loosen their own reply and chat settings. Their posts, likes and follows are public like everyone's. And adult content that nobody labeled isn't hidden.
Does Bluesky Moderation Actually Catch Bad Accounts?
It catches some of them fast. Bluesky doesn't publish a live count of what moderation does to everyday posts, so I took one hour of posts and checked what had happened to them a day later.
| What happened to the post after ~25 hours | Posts | Share of 25,000 |
|---|---|---|
| Still up | 24,249 | 97.0% |
| Gone: account taken down | 693 | 2.8% |
| Gone: account deactivated by its owner | 32 | 0.1% |
| Gone: post removed, account still active | 26 | 0.1% |
Three things in that table matter:
- 57 accounts made all 693 of the posts that vanished with their accounts. That's 0.3% of the 18,463 accounts in the sample, averaging about 12 posts each in one hour against 1.4 for the sample as a whole. One of them alone wrote 319 of my 25,000 posts, about one in 80. All of it was gone when I checked, about 25 hours later.
- Moderation does most of the labeling. Of the 24,249 posts still up, 464 (1.9%) carried an adult-content label (porn, sexual, nudity or graphic media). Bluesky's moderation added the label on 320 of them. 272 (59%) were flagged only because moderation labeled them; the poster hadn't.
- Adult content is concentrated in media. Among posts with images or video, 9.2% (415 of 4,498) carried an adult-content label. Among the rest, 0.25% did.
For scale, Bluesky's 2025 report says users filed 9.97 million reports, Bluesky applied 16.49 million labels (14.2 million for adult content), and it took down 2.08 million accounts that year. For who's behind the automated accounts, see my census of bots on Bluesky.
Is Bluesky Safer Than X?
For what lands in your feed, usually yes; for privacy, no. On Bluesky, porn and graphic content is hidden unless you opt in, the Following feed is plain chronological, and you can subscribe to shared block lists and extra moderation services run by other people. X lets you protect your account so only approved followers see your posts, and makes likes private. Bluesky has neither. So Bluesky is the better pick if you want control over what you see, and the worse one if you want control over who sees you.
Is Bluesky Safe From Hackers and Scams?
Most Bluesky account takeovers start with phishing, not a hack of Bluesky itself. On October 17, 2026, Bluesky's Safety team said a phishing attack had "compromised more than 4,500 accounts", spread through posts about a fake government payment. Four habits cover most of it:
- Turn on two-factor authentication in Settings > Privacy and security. It sends a sign-in code to your email.
- Only type your password into the Bluesky app, bsky.app or bsky.social.
- Ignore DMs about your account. Bluesky's Safety team says it will "NEVER DM you about account verification", that "DMs about account violations or moderation issues are scams", and that it tells you about moderation decisions by email.
- Give third-party apps an app password or Bluesky's own login screen (OAuth), never your main password. You can revoke either one later.
Bluesky's official accounts mostly have handles ending in bsky.app, bsky.team or bsky.network, plus a blue check. I tracked the October attack and the fake "Bluesky Support" accounts in Bluesky scams.
Is Bluesky Safe From AI Training?
From Bluesky itself, yes. From everyone else, no. In November 2024 Bluesky posted: "We do not use any of your content to train generative AI, and have no intention of doing so." But posts are public, so outside crawlers can read them. That same month, someone posted a dataset of a million Bluesky posts on Hugging Face (it was taken down after complaints).
A March 2025 proposal, "User Intents for Data Reuse", would let you tell other companies not to train on your posts. As of November 16, 2026 it isn't a setting in the Bluesky app, and it would only ever be a request, not a lock.
How I checked. I replayed one hour of Bluesky's public Jetstream firehose (16:00 to 17:00 UTC, November 15, 2026): 164,168 new posts and replies in every language. I picked 25,000 at random and looked each one up through Bluesky's public API about 25 hours later, which also returns any labels from Bluesky's own moderation and the poster. For each missing post, I asked the network's relay for its account's status (active, taken down or deactivated). A post missing from an active account could be deleted by its author or removed on its own by moderation; the API doesn't say which. The 22.1% figure counts sampled posts whose author carries the logged-out opt-out label. This is one hour of one day, so read it as what moderation caught in that hour. It can't see harassment, which is about who replies to whom, and it can't count bad posts nobody labeled. App defaults are from Bluesky's open-source app code as of November 16, 2026.
The advice above applies to growth tools too: give them OAuth or an app password, never your main password. I built Agent Sky that way. It connects through Bluesky's own login screen or an app password, follows active people in your niche at a pace you set, and checks for a real follow-back before it unfollows anyone. It's $9 a month and free to start.
Quick FAQ
Is Bluesky safe for kids and teens? Bluesky's terms require users to be at least 13 and to meet any higher minimum age in local law. Under-16s are blocked in Australia and Virginia, and under-18s in Tennessee and Mississippi. Porn, sexual and graphic content is hidden by default and can't be turned on with a declared age under 18, and DMs default to people you follow. But there are no parental controls, a teen can change their own settings, and everything they post, like or follow is public.
Is Bluesky safer than X? For not seeing content you didn't ask for, usually yes. Adult content is hidden by default, the Following feed is chronological, and you can subscribe to shared block lists and extra moderation services. For privacy, no: Bluesky has no protected or private accounts, and likes, follows and blocks are all public.
Is Bluesky safe from AI scraping? Bluesky the company said in November 2024 that it does not use your content to train generative AI and has no plans to. But posts are public, so outside companies can still collect them. A March 2025 proposal would let users signal whether they consent to AI training, but it wasn't a setting in the Bluesky app as of November 16, 2026, and it would only be a request.
Is Bluesky safe from hackers? The service itself is fine; phishing is what gets people. In October 2026 a phishing attack compromised more than 4,500 Bluesky accounts. Turn on two-factor authentication in Settings > Privacy and security, only enter your password in the Bluesky app, bsky.app or bsky.social, and give third-party tools an app password or OAuth login, never your main password.