Back when the first moat creation gambit for AI failed (that they were creating SkyNet so the government needs to block anyone else from working on SkyNet since only OpenAI can be trusted to control it not just any rando) they moved onto the safety angle with the same idea. I recall seeing an infographic that all the major players were signed onto some kind of safety pledge, Meta, OpenAI, Microsoft, etc. Basically they didn't want anyone else training on the whole world's data because only they could be trusted to not do nefarious things with it. The infographic had a statement about not training on CSAM and revenge porn and the like but the corpospeak it was worded in made it sound like they were promising not to do it anymore, not that they never did.
I've tried to find this graphic against several times over the years but it's either been scrubbed from the internet or I just can't remember enough details to find it. Amusingly, it only just occurred to me that maybe I should ask ChatGPT to help me find it.
> The infographic had a statement about not training on CSAM and revenge porn and the like but the corpospeak it was worded in made it sound like they were promising not to do it anymore, not that they never did.
We know they did, an earlier version of the LAION dataset was found to contain CSAM after everyone had already trained their image generation models on it.
I've tried to find this graphic against several times over the years but it's either been scrubbed from the internet or I just can't remember enough details to find it. Amusingly, it only just occurred to me that maybe I should ask ChatGPT to help me find it.