Season 7 has been the most balanced game season yet, and for this reason, we’re not going to see many changes. Unless these minor changes are breakind and will bring some boring/stale meta, it’s unlikely that the following seasons will see major changes as well.
You can read the patch details here:
The main difference is that, unfortunately, the ChatGPT-powered bots will be disabled on November 1st, 2023 after Valve explicitly requested their removal.
If you are not familiar with Untrusted or why they are a “big-deal” for our small community, you can see a short video here about them:
Wait, what happened?
It’s no secret it’s often hard to get enough players to kick-off a lobby, and once a game is going it’s a lot easier to get people in, so having LLM-powered bots certainly made it nicer to cut waiting times a little bit.
Some of you may have noticed that after the chat-gpt powered bots were released on untrusted, I had started improving them a little bit, as part of my maintenance effort of Untrusted, and then the updates suddenly stopped.
Long story short, after an initial inquiry, I’ve been asked by Valve via their official ticketing support system to remove chatGPT from Untrusted or it would get pulled off Steam. This happened on September 18th. Since then, I’ve asked multiple times if I was allowed to share the exact wording they used (please remember that Steamworks developers have to abide to an NDA) but haven’t gotten a reply yet.
Until that happens (if that happens), I can just only say that the issue here is with their policy, point 5 that states that you should not publish content you don’t own or have adequate rights to.
Generative-AI is changing the landscape and it’s understandable their lawyers are taking the safe route.
Even if the ChatGPT bot is running on servers owned by myself and ChatGPT is never called directly by the users, which I believe is an important distinction (and valve has been made aware of this fact), the request still was made.
In my personal and biased opinion, going against all generative-AI made content with a blanket ban is not the right solution. I don’t think openai is breaking copyright (which I really doubt, but I’m sure we’re going to discover that in the upcoming months or years) and I don’t blame Valve for going the safe route, but I would have appreciated a faster back and forth in the ticketing system in the absence of a clear policy, if anything.
So what now?
I had two choices at this point:
Choice A: split the web and steam community, and keep the bots only for web players, breaking cross-compatibility (As a reminder, you can play for free on itch or kongregate, but the Steam version gives you a little bit more goodies and is overall a better experience)
Choice B: turn off the bots.
From now on…
I’m turning off the bots on November 1st, and hopefully Valve will set a clearer, more permissive, policy on generative AI which will allow me to turn the bots back on. Though to be honest, this experience entirely put me off the development of bots for untrusted.
I’d like to point out that Valve hasn’t set a hard cutoff date, but asked generally what my plans were. I’ve shared those with them, and I’m going through with those.