The social network has long been working on encrypted messages, which ensure that messages cannot be intercepted while they are being delivered. That is already possible on most other major messaging platforms, such as WhatsApp and iMessage.
Now Twitter has released what it says is the first step towards integrating that security technology into its own product. But it says that it is not fully safe, and that users should not rely on it to keep messages secure.
“Early version of encrypted direct messages just launched,” Mr Musk wrote on Twitter. “Try it, but don’t trust it yet.”
On its support pages, Twitter’s aim is that “if someone puts a gun to our heads, we still can’t access your messages”. But it says it is “not quite there yet” and still “working on it”.
That suggests that the messages are not end-to-end encrypted. While messages are secured from some attackers, anyone at Twitter or with access to its servers may still be able to access them.
What’s more, Twitter’s implementation comes with a host of other limitations. It is only available when both of the users are verified and pay for Twitter’s $8 a month service, and users must turn it on rather than it being enabled by default.
The messages also cannot be sent to groups, include pictures or other attachments, cannot be joined by new devices, are only supported on a limited number of devices per account and do not have proper protection against certain cyber attacks.
When those various conditions are satisfied, users will be informed they are having an encrypted conversation by a message that appears within the chat. The other user’s profile picture will also show a lock icon over it.
Twitter said that it would be updating the feature to remove some of those limitations, and would provide more information about the technology underpinning it in releases later this year.