In early January of 2026, the communications tech giant Discord posted a survey regarding various types of AI integration on their platform. The questions posed included several invasive monitoring ideas, such as analysis of audio and video chats, seemingly in real time, with no distinction for private servers—or direct messages.
Seemingly due to overwhelming backlash, the survey was taken down a mere several hours after it initially launched. In the free form section at the end, I wrote a couple paragraphs airing my concerns. While I've entirely moved to Matrix now, I still felt it worthwhile to respond, and also to mirror this text elsewhere.
As a large, centralized corporation housing an enormous amount of personal data of users—including that of minors—I am incredibly concerned with any AI integration into Discord whatsoever. The thought of any person having the ability to offload data, particularly from private servers or private messages, into some other company's data centers for processing, is a privacy nightmare. With some of the features considered within this very survey, it sounds like even voice and video chats may become rife with direct processing to such companies. I cannot imagine whoever you partnered with would sit idly by while such massive natural human data streams come in, without harvesting them for who knows what.
Even if you hypothetically added an opt-out, and I were to trust that this genuinely prevented the transfer of data, the vast majority of people will leave settings as default—having the effects of this without necessarily even realizing the ramifications. It may also violate the consent of others, dependent on if someone who chooses to have AI highlights or summaries, have information included from participants who have chosen to opt-out. This would defeat the entire purpose of turning it off, beyond a veneer of a few hidden buttons; this would obviously be very problematic.
Additionally, I fundamentally am tied to Discord for communication between people, and only people. AI generated stickers, emoji, messages, or any other sort of non-human output, are directly at odds with the only thing I care about doing on the platform: talking to actual people. Groups of friends or communities create their own in-jokes and memes, and it is this human interaction which spurs the creation of things which we collectively enjoy. When I use a custom emoji or sticker, there is a story as to why it's there, and who created it. If there are people who wish to use generative models in spite of that, I would encourage them to use a local only, open weight, generative model they have full control over on their own computer.
I certainly don't want to further platform and normalize the generative models of the multi-trillion dollar AI market bubble, which is almost certainly what would be the case for any AI integration that is done by Discord. This bubble is causing lots of problems currently, and I expect those problems to get worse before they get better. Look no further than the gaming demographic Discord used to be aimed at; SSD and RAM prices are at astronomical levels right now, purely because of backroom AI data center deals. I am appalled by the prospect of further rewarding this behavior as it continues to put strain on the general populace, with deep pocket investors begging us to take part in their self-serving, prospective profiteering.
All writing is licensed under CC-BY-ND with attribution to Foxfire. The FSF recommends this license for opinion pieces.