Read this before you rely on it
Tox is encrypted. Tox is not anonymous.
Plenty of sites about Tox — including an earlier version of this one — list "anonymity" as a feature. That is wrong, and getting it wrong is how people end up trusting a tool with something it was never built to protect. Tox encrypts your conversations very well. It does not hide where you are.
The short version
In a peer-to-peer network, packets travel directly between two machines. For that to work, each machine has to know the other's address. When you accept a friend request, that person can see your IP address. So can anyone who compromises their device, and so can anyone who successfully sends you a friend request you accept.
If your safety depends on nobody learning your location or your internet provider, Tox on its own is not enough. Route it through Tor, or use a tool designed for that threat model.
What Tox does protect
- Message content. End-to-end encrypted with NaCl — Curve25519 for key exchange, XSalsa20 for the payload, Poly1305 for authentication. Always on, no way to disable it, no plaintext fallback.
- Your identity from the phone system. No phone number, no email address, no real name, no account with any company. You generate a keypair on your own machine and that is your identity.
- Your history from seizure. There is no server holding your conversations, so there is nothing to subpoena, breach or hand over. Your history lives on your own device and your contacts' devices, and nowhere else.
- Message integrity. Poly1305 authentication means a tampered packet is rejected, not delivered. Nobody in the middle can alter what you said.
- Metadata from a central operator. No company builds a graph of who talks to whom, because no company is in the path.
What Tox does not protect
Your IP address from contacts
The big one. Direct connections require addresses. Everyone on your contact list can see roughly where you are and who your internet provider is. Adding a stranger means giving a stranger your IP.
The fact that you use Tox
Your provider and network administrator cannot read your traffic, but they can see its shape. Tox is identifiable by deep packet inspection and is blocked on that basis in several countries.
Your DHT activity from nodes
Any bootstrap node — including this one — sees the IP addresses that query it and when. Most clients rotate their DHT key at every start, which limits correlation, but does not eliminate it.
Anything on a compromised device
Encryption ends where the screen begins. Malware, a keylogger, or somebody with your unlocked laptop reads everything. No protocol fixes this.
Which threat model are you in?
| You want to… | Tox alone |
|---|---|
| Keep conversations away from advertisers and data brokers | Works well |
| Message without giving anyone a phone number | Works well |
| Avoid a company that could be breached or subpoenaed | Works well |
| Talk to people you already know and trust | Works well |
| Hide your location from the people you are talking to | Needs Tor |
| Hide from a network operator that you use Tox at all | Needs Tor or a bridge |
| Protect a source, or organise under a hostile government | Not sufficient on its own |
Running Tox over Tor
Tor hides your IP address from the people you talk to; Tox encrypts what you say to them. Together they cover more than either does alone.
The mechanics: Tor carries TCP only, so you must disable UDP in your client and force TCP-only mode. That is not a small change. With UDP off, every connection goes through TCP relays like this node instead of directly to your peer, so expect noticeably higher latency and slower file transfers. Audio and video calls generally become unusable — real-time media does not survive the added delay.
Set your client's proxy to your local Tor SOCKS5 listener, usually 127.0.0.1 on port 9050, then enable TCP-only mode in the same settings section. Verify it worked before trusting it: if the client connects with the proxy pointed somewhere invalid, it is not using the proxy.
One caveat worth stating plainly: this configuration is fiddly, and a misconfiguration fails open — you stay connected while leaking your real address. If your safety genuinely depends on this, test it with a machine you control on the other end before you rely on it.
What this node logs
This node runs tox-bootstrapd in its standard configuration. It answers DHT queries and relays TCP. It has no key that can decrypt anything passing through it, and no plaintext ever exists on this machine to be logged.
What it does necessarily process, as any node must: the IP addresses that connect to it, their DHT public keys, and connection timing. Treat any public node — this one included — as a party that can observe when you were online and from where. That is not a reason to avoid public nodes; it is a reason to know what you are trusting them with. Details are on the status page.
If you need more than this
Different tools make different trades, and the right answer depends on what you are protecting against:
- Signal — strong encryption and metadata protection, but requires a phone number and routes through central servers.
- SimpleX Chat — no persistent identifiers at all, self-hostable relays, designed specifically around metadata resistance.
- Briar — built for hostile network conditions; routes over Tor by default and works peer-to-peer over Bluetooth or local Wi-Fi when the internet is cut off.
- Matrix / Element — federated rather than peer-to-peer, so your IP goes to your homeserver instead of your contacts.
Tox's distinctive strength is that it is genuinely serverless and genuinely leaderless. That is worth something real. It just isn't anonymity.