What is HAL?

Currently in private beta, HAL (@thisishalbot) is an anti-spam bot that uses federation partners and expression filters to fight spam. It also has an anti-duplication feature, and has the ability to block instant-replies to channel posts — 100% of the time, those are spam. HAL can clean service messages and bot commands, too. It is configurable to a certain degree, with more features and configurability coming later down the line.

Is HAL free?

In private and public betas, yes, HAL will be free. Additionally, HAL’s anti-spam feature will always be free. Later down the line, HAL may acquire some group management features, and we may choose to charge for that. However, it’s early days. Lots of work to still do, and lots to think about.

If you would like to join the private beta, drop us a note in the support group and we’ll take it from there.

Can HAL be white-labelled?

No. At least, not yet. We haven’t really thought about it. However, HAL may well introduce a REST API that can be used by custom bots to check for spammers. Portions of the API would be free, others would be paid, but we haven’t given it much thought yet.

Where can I get help using HAL?

First, check the /help menu in a private chat with the bot. If you can’t find what you’re looking for, then you can reach out in the support group.

I’ve been banned by HAL unfairly. How do I get unbanned?

You can send your appeal to the support group and an admin will assist you, subject to any procedures outlined there.

Support Group: @halsupport

#bugfix

When using a ban command in a channel discussion thread without replying to a comment in that thread, HAL will no longer ban the channel and delete auto-forwarded channel-posts for the past 48-hours. This was more of an oversight than it was a bug – Telegram works in strange ways. 🤷‍♂️

#internals

HAL now has baked-in rate-limiting to help keep server resources available when under load. Also, for cluster actions (such as active banning), outbound request throttling has been activated.

Update: this internal change may be adjusted from time to time so we can find a nice balance between performance and prevention of flood-waits from Telegram’s servers.

#feature

Comment moderation has been added to HAL. Available settings are off, light, moderate and heavy. See “Moderation” in /help for more information.

Note: this feature may move to HAL Premium as a paid feature

#feature

The ability to delete private links in messages is now available. To use it, send /deleteprivateinvitelinks on to your group, or a connected DM with HAL. See the “Antispam” section in /help to learn more.

Note: Though unconfirmed, this feature may move to a “locks” feature in the future, and that feature may only be available to HAL Premium.

#update

HAL will no longer ban users who were banned by SpamWatch for spam-adding members to groups. Our simple rule is that we only ban for actual spam posted in a group, or for impersonation/unsolicited messages. These bans have resulted in false-positives on genuine users who had no intent on truly spamming.

⚠️ #incident

One of our federation partners, CAS, is currently unreachable by HAL. Due to a bug, this has unfortunately caused HAL to fall over and not process any messages.

We’ve found the bug and fixed it, so messages should still at least be subject to analysis through the filtering system, however CAS bans currently have no effect.

We’ll post another update here once we’ve heard back from CAS, or once it comes back online.

✅ #incident resolved at ~09:20 UTC

The CAS API is now back online, so HAL will continue to check for CAS bans on all new messages.

Apologies for the inconvenience.

⚠️ #incident

SpamWatch went down several times for system updates, and once it came back online, we noticed that all SpamWatch checks were resulting in false-positives, which banned several users globally in error. These users have been unbanned, and SpamWatch ban-checking has been disabled while we investigate.