AI safety is fear mongering to shut up the Luddites
The AI we have now (Stable Diffusion, chatgpt) are technical advancements that allow inferior but cheaper production of artistic content. It is not a step closer to death-by-paperclips; it is merely another step of big capital automating production, hoarding more wealth in a smaller group.
The closer thing to AI safety is unsupervised execution of laws by ML.
Stable Diffusion, where the best models are freely downloadable and the company behind them is losing money and currently doesn't look like it has much of a future, is your example of "big capital automating production, hoarding more wealth"?
At least with OpenAI/GPT, the models are actually kept behind a door, though there are still downloadable competitors that keep insisting they're not terrible :P
I'd say stability AI directly are not doing that very well, but the vast majority of users of image generation AI are using NovelAI or Midjourney, both of which are raking in cash and don't share their models (NovelAI seems to be at least mostly finetunes of SD, while there's not much public details on midjourney's models at all). Both cases don't have much 'big capital' (especially compared to OpenAI) though: they're just very profitable from getting on the train of making it easy to use early.
> Stable Diffusion, where the best models are freely downloadable
Stable Diffusion 1.6, Stable Diffusion 3 Large & Large Turbo, the model(s) — originally an SDXL Turbo-based finetune, but explicitly stated to be planned to evolve over time and potentially eventually use multiple models — used for the Stable Image Core service are all not publicly released under any license, and SD3 Medium under only a restrictive license.
In the places I see people working with them, 3 is considered a disappointment (though noisy people are often those with the negative opinions, so take with a pinch of salt), while most of the interesting stuff is either SDXL or what third parties have done by fine-tuning on 1.5 or 2.
The AI we have now (Stable Diffusion, chatgpt) are technical advancements that allow inferior but cheaper production of artistic content. It is not a step closer to death-by-paperclips; it is merely another step of big capital automating production, hoarding more wealth in a smaller group.
The closer thing to AI safety is unsupervised execution of laws by ML.