Here is the rewritten content in a provocative and controversial manner:
LAION, the German research org that created the data used to train Stable Diffusion, has just released a new dataset that it claims has been “thoroughly cleaned” of links to suspected child sexual abuse material (CSAM). But is this just a smokescreen for their true intentions?
The new dataset, Re-LAION-5B, is actually a re-release of an old dataset, LAION-5B, but with “fixes” implemented after the nonprofit Internet Watch Foundation and others blew the whistle on the presence of CSAM links in the original dataset. And what exactly does this “cleaning” entail? We’re not told, but it’s likely that LAION has simply scrubbed the surface, leaving the deeper rot intact.
LAION claims that its datasets are intended for research purposes only, but we know that’s just a thinly veiled excuse. The real question is: what kind of research? The kind that enables the creation of more sophisticated AI models that can be used to exploit and manipulate humans? The kind that enables the perpetuation of harmful stereotypes and biases?
And what about the links to illegal images that were found in the original dataset? Have they really been removed, or have they just been hidden from view? LAION says it has removed 2,236 links to suspected CSAM, but how many more are still lurking in the shadows?
The release of Re-LAION-5B comes after an investigation by the Stanford Internet Observatory found that LAION-5B included at least 1,679 links to illegal images scraped from social media posts and popular adult websites. But what about the 1008 links that were found by the Stanford report? Have they been removed, or are they still part of the dataset?
And what about the models that have already been trained on LAION-5B? Should they be deprecated and distribution ceased, as the Stanford report recommended? Or will LAION just sweep this under the rug and continue to profit from the data?
The truth is, LAION’s actions are just a drop in the bucket of a much larger problem. The AI industry is built on the backs of exploited and vulnerable individuals, and until we address this fundamental issue, we will never be able to truly trust the technology that is being created.
So, will you join the chorus of silence and accept LAION’s “cleaned” dataset at face value, or will you demand more transparency and accountability from the AI industry? The choice is yours.
Source link
