The TASVideos site code currently contains a very small amount of human-reviewed agentic code, committed before this policy was drafted. In accordance with this policy, however, no agentic code will be used by any site developer going forward.

TASVideos does not support LLM usage in its current state. The environmental impact caused by data centers, the emotional and mental damage done to people manipulated to rely on it, and the overall ethical concerns of, among many other things, harvesting data from non-consenting users and not only selling it back to them, but allowing others to effortlessly profit from it, are far too much to overlook individually, let alone all together.

This policy will not change until the AI industry does. For as long as it remains unchallenged, unregulated, and unethical, TASVideos staff and developers will remain opposed to using it.

Unfortunately, we are forced to live in a reality where LLMs are rapidly proliferating and forcibly being implemented into every facet of modern technology, and thus we cannot fully guarantee that no LLM usage will ever appear on the site. As a large community, one that both relies on many pieces of outside software and relies on user-submitted content in order to function, it is sadly inevitable that something outside of our control that we support will adopt LLM assistance, or already has. With the site code being open source as well, we cannot even guarantee the site itself will remain free of agentic code usage going forward. We will be doing our best to catch and prevent everything we can, but we may and most likely will slip up and make mistakes.

We also recognize that LLMs are a tool, and like any other tool it is possible to use them for beneficial purposes, particularly in accessibility. Some users, such as those who are disabled or those who do not speak or understand English, may rely on LLM text generation in order to be able to communicate on the site at all. As much as we wish these tools could be used in much better circumstances, we have chosen to allow them as a way to ensure accessibility to the site is as open as possible.

That being said... A person's own words will always be preferred over an LLM's output, no matter how they might sound. We encourage any sufficiently abled user to write in English to the best of their ability, regardless of their fluency. There is no such thing as "broken English", after all. For those who cannot write in English, but can write in their native language, we encourage that as well, as we have a number of users in the community who are willing and able to translate.

TASVideos forbids the posting of fully or majority LLM-generated media such as images, videos, and music. Users posting such material will be warned, and the material will be removed. Repeated offenses, or those that we deem to be severe or contextually hostile, will be escalated to temporary or indefinite bans.

Media containing AI-generated content is not strictly forbidden, though we ask that it be explicitly labelled and presented as such within the media itself, and that said media is not an explicit endorsement of LLM usage. TASVideos reserves the right to remove anything we see fit in accordance to this.

Users posting AI-generated media with the intention of profiting from it will be indefinitely banned without warning or hesitation.

Discussion of AI in any form is still allowed, though it is of course still subject to moderation in accordance with the site rules. We will not moderate users simply for supporting AI, unless they make it clear that is their only reason for being here.

We would like to ask Discord users to only discuss AI in the thread in #general so that those wishing to avoid it completely aren't forced to see it.

TASVideos is a gaming community, so the term "AI" is and will continue to be used frequently in multiple different contexts. Unless explicitly stated otherwise, these usages of AI as a term are not to be confused with the modern colloquial usage of AI in reference to LLMs.

Severity should be obvious enough for both us as staff and for the rest of the community at large to inherently agree on. "Contextually hostile" is, as said in an earlier clarification, if we deem a user's reason for being here is explicitly to support or defend LLMs or the AI industry at large.

Machine learning algorithms and neural networks, while they may be referred to as generative AI, are still perfectly fine to explore and post, as the technology for them has existed for far longer and is far less exploitative than modern day LLMs.

The reason we ask for your own words is to prevent misinformation, particularly with submissions. Any information that you want or need to share with us could easily be mistranslated, changed, or even outright lost when passed through an LLM. Inaccurate information could potentially disrupt site operations by making submissions harder to process efficiently, and given how niche TASVideos is as a site and TASing is a hobby, the chance of an LLM producing inaccurate information about it is effectively guaranteed.

For those of you who wish to completely avoid AI-powered software, here are some curated collections:

https://codeberg.org/brib/slopfree-software-index
https://noai.starlightnet.work/list.html