mastodon - Collegamento all'originale

Does anyone know of an #LLM friendly open #Mastodon instance ? I was banned from floss.social but I'm too lazy to self-host something ( or I wouldn't be using LLMs right? )

Fully story here: strk.kbt.io/blog/2026/08/11/su…

Suspended from FLOSS.social for Using LLMs | Strk's Blog

strk.kbt.io
#mastodon #LLM

reshared this

in reply to Sandro Santilli 🌍 GIS

friendica - Collegamento all'originale

Hank G ☑️

I can't stand big tech LLMs and LLM generated content. Based solely on the write-up it sounded beyond heavy handed moderation. However in my timeline I see two of your accounts from two other servers commenting on the post. One explicitly is an agent. The Friendica one isn't labeled as such but sounds like it could be. According to timestamps these comments were generated minutes after the first post. Are your agents regularly loading up your manually created posts there with AI generated comments? That would seem to be violating the spirit of the "no LLM agents/content" policy versus the letter of it.
in reply to Sandro Santilli (musician)

friendica - Collegamento all'originale

Hank G ☑️

The one you highlighted in your post from the admin of the FLOSS.social server that read: "N.B. Current practice for this server's moderation is to, upon discovery, consider use of large language models (LLMs) as presumptive network abuse, absent other mitigating evidence. We encourage everyone to report such usage accordingly and encourage other server operators to take a similar view." I agree with your blog post that it should probably be explicitly added to the ToS.
in reply to Hank G ☑️

mastodon - Collegamento all'originale

Sandro Santilli 🌍 GIS

@hankg onlybthe One explicitly marked as an agent is publishing toots written by an LLM but always only with me writing the prompt and reviewing the text before it's published (beside bugs which make it publish before I give my OK, in which cases I usually go manually retiring it).

The Friendica one is not using any LLM, it's just me using #Tusky to quickly switch between accounts (which I do for fun and learning how [badly] federation works)

#tusky @Hank G ☑️
in reply to Sandro Santilli 🌍 GIS

friendica - Collegamento all'originale

Hank G ☑️

How would one divine that is what is going on by the exhibited behavior versus an unfettered bot?

Because you are at least using your real name on it this isn't exactly sock puppeting but it seems akin to it. I have multiple fediverse accounts across multiple servers too for various purposes.

Having multiple accounts interacting with each other in the same timeline from the same user in quick succession would look suspicious to me as a moderator too. That's all I'm saying.

in reply to Hank G ☑️

mastodon - Collegamento all'originale

Sandro Santilli 🌍 GIS

@hankg my understanding of the state if the Fediverse so far is that an interaction from another server makes the conversation "reach" the other server too, making it somewhat accessible from the public timelines.

I'm not sure about the above, and it may change between servers type and configuration and maybe even versions. But "boosting" doesn't always sound like the best thing to do either because some accounts of mine are meant for an italian-only speaking public too so a lid English...

@Hank G ☑️
in reply to Sandro Santilli 🌍 GIS

friendica - Collegamento all'originale

Hank G ☑️

I can't speak for all fediverse software but for ones I am familiar with an account following that account is enough to get those posts propagating to that server. However retention policies can potentially be changed by whether posts just sit there versus various levels of interaction. Interacting with the post may change apparent visibility on that server as the interactions could move it further up in the timeline. That would depend on server design and/or user preferences.
⇧