LAION, the German analysis org that created the information used to coach Stable Diffusion, amongst different generative AI fashions, has launched a brand new dataset that it claims has been “completely cleaned of identified hyperlinks to suspected youngster sexual abuse materials (CSAM).”
The new dataset, Re-LAION-5B, is definitely a re-release of an outdated dataset, LAION-5B — however with “fixes” carried out with suggestions from the nonprofit Internet Watch Foundation, Human Rights Watch, the Canadian Center for Child Protection and the now-defunct Stanford Internet Observatory. It’s obtainable for obtain in two variations, Re-LAION-5B Research and Re-LAION-5B Research-Safe (which additionally removes further NSFW content material), each of which have been filtered for 1000’s of hyperlinks to identified — and “doubtless” — CSAM, LAION says.
“LAION has been dedicated to eradicating unlawful content material from its datasets from the very starting and has carried out acceptable measures to attain this from the outset,” LAION wrote in a weblog submit. “LAION strictly adheres to the precept that unlawful content material is eliminated ASAP after it turns into identified.”
Important to notice is that LAION’s datasets don’t — and by no means did — comprise photographs. Rather, they’re indexes of hyperlinks to pictures and picture alt textual content that LAION curated, all of which got here from a completely different dataset — the Common Crawl — of scraped websites and net pages.
The launch of Re-LAION-5B comes after an investigation in December 2023 by the Stanford Internet Observatory that discovered that LAION-5B — particularly a subset referred to as LAION-5B 400M — included at the very least 1,679 hyperlinks to unlawful photographs scraped from social media posts and common grownup web sites. According to the report, 400M additionally contained hyperlinks to “a variety of inappropriate content material together with pornographic imagery, racist slurs, and dangerous social stereotypes.”
While the Stanford co-authors of the report famous that it could be troublesome to take away the offending content material and that the presence of CSAM doesn’t essentially affect the output of fashions educated on the dataset, LAION stated it could briefly take LAION-5B offline.
The Stanford report really helpful that fashions educated on LAION-5B “must be deprecated and distribution ceased the place possible.” Perhaps relatedly, AI startup Runway not too long ago took down its Stable Diffusion 1.5 mannequin from the AI internet hosting platform Hugging Face; we’ve reached out to the corporate for extra data. (Runway in 2023 partnered with Stability AI, the corporate behind Stable Diffusion, to assist practice the unique Stable Diffusion mannequin.)
Of the brand new Re-LAION-5B dataset, which comprises round 5.5 billion text-image pairs and was launched underneath an Apache 2.0 license, LAION says that the metadata can be utilized by third events to wash present copies of LAION-5B by eradicating the matching unlawful content material.
LAION stresses that its datasets are supposed for analysis — not industrial — functions. But, if historical past is any indication, that received’t dissuade some organizations. Beyond Stability AI, Google as soon as used LAION datasets to coach its image-generating fashions.
“In all, 2,236 hyperlinks [to suspected CSAM] have been eliminated after matching with the lists of hyperlink and picture hashes offered by our companions,” LAION continued within the submit. “These hyperlinks additionally subsume 1008 hyperlinks discovered by the Stanford Internet Observatory report in December 2023 … We strongly urge all analysis labs and organizations who nonetheless make use of outdated LAION-5B emigrate to Re-LAION-5B datasets as quickly as doable.”
