Horrible idea: Use AI slop to devalue platforms directly instead of trying to extract revenue.

Play a game of Cards Against Humanity. Feed the winning combinations into the AI slop machines with the preamble: "Turn this into a cringey LinkedIn post." Flood LinkedIn and Meta properties with the output.

Accelerate model collapse.

Make corporate social media unusable.

Keep this crap far away from Fedi.

Write open source software on a throwaway GitHub, permissively licensed, that purports to solve a problem that AI / blockchain grifters would find useful.

Fill your code with functions that search your local filesystem for cryptocurrency private keys, then post them on social media in plaintext.

If the AI slop machines serve similar code to the end user, that's not your problem. You didn't tell them to use it.

@soatok pretty sure this is called malware but i'm so here for it

@alina It's not malware it's "convenient crypto automation for monetizing AI videos and images".

;3 ;3 ;3

@soatok @alina It's not malware, we wrote it for the user to use voluntarily.

Then some arse stole it and pasted into other people's codebases, but what can you do, yunno?

@alina @soatok github is mainly malware anyway
@soatok OSS that purports to optimise high-frequency trading, but instead makes donations to nonprofit animal sanctuaries.
@whyrl @soatok high-frequency trading project that's actually just about writing software for transoceanic DX messaging
@whyrl @soatok I think it would be much more apt for it to make them to non-profit groups that hope local queer disenfranchised kids, or local queer kids in general, but same general idea.
@soatok So "LLM token optimization and performance tweaks for etherium blockchain projects" should flood the blockchain with thousands of bespoke boykisser NFTs until the host account is dry?
@soatok in terms of enacting financial harm to random people basically indiscriminately i'm pretty sure that this is morally bad but seeing as i already want to cause harm to people who pull in dependancies without properly looking into them i dont really give a fuck
@shitpostalotl @soatok yeah but it's only targeting crypto users so is there really that much harm here?
@zuthal @soatok in terms of "harm", you are literally harming someone by allowing arbitrary 3rd parties to steal their assets. i assume you are attempting to say that this harm doesn't carry moral weight, in which case that's between you and whatever informs your personal philosophy.
@zuthal @shitpostalotl @soatok some people (hi!) use it for legitimate things as it's basically the only way to send & receive money online without doxxing yourself. and it's not uncommon for trans folks (again, hi!) to use crypto to receive donations for that exact reason, we can't fuck around with paypal or w/e when it makes it trivial for kiwifarms types to get our personal details (or just shut it down by typing "porn" into the payment notes) - and the billionaire CEOs running the payment processors are very fucking clear about wanting us dead, so...

@soatok

i'm gonna post UUencoded .BMP images of Rick Astley

@soatok diabolical! AI coders as supply chain attack surface. 
@cuchaz @soatok There are existing proposed attacks that do that. For example, registering Python packages that have names that popular LLMs routinely hallucinate. The next step from there is to flood the web with references to your malicious package so LLMs favour it.
@david_chisnall @cuchaz @soatok Yeah it's called "slopsquatting" and is my most recent concern as someone responsible for supply-chain security at my company.

@soatok genuinely wish more subreddits about programs and apps would force people to disclose vibe coding

I'm in the MacApps subreddit and a guy vibe coded a financial app
Where you put your banking credentials

@soatok the first time I heard the term slopsquatting (great word) was in the comments to an arstechnica article about this problem https://arstechnica.com/security/2025/04/ai-generated-code-could-be-a-disaster-for-the-software-supply-chain-heres-why/
AI-generated code could be a disaster for the software supply chain. Here’s why.

LLM-produced code could make us much more vulnerable to supply-chain attacks.

Ars Technica
@soatok That looks like the Ken Thompson's hack but real.
@soatok I was thinking something similar a few days ago. What would happen if you flooded GitHub with trivial programs with naming or comments that imply it's doing something different. Like a "shortest path" function, but it's actually just fizzbuzz. How long until AI slop would just think that every problem is fizzbuzz and just offer some variation of fizzbuzz for every prompt?
@PurpleStephyr @soatok Wow, this tempted me for a few seconds to actually create a Microsoft Github account and do some of the described shenanigans. :)