Thanks to Bluesky’s open API, anyone can evaluate your data for AI training

Bluesky may not train AI systems on user content like other social networks do, but there’s little to stop third parties from doing so.

According to a report by 404 Media, a machine learning librarian at AI company Hugging Face retrieved 1 million public posts from Bluesky for machine learning research through its Firehose API and transferred the dataset to a public repository. Daniel van Strien later removed the data due to the resulting controversy. However, they serve as a timely reminder that anything you post publicly on Bluesky is public.

Bluesky said it is looking at ways to give users the ability to communicate their consent preferences externally, leaving it up to the parties whether to respect those preferences.

The company wrote: “Bluesky cannot enforce this consent outside of our systems. It is up to external developers to respect these settings. We are in ongoing discussions with engineers and lawyers and hope to provide further updates on this topic soon!”

What is clear here is that while Bluesky is becoming increasingly popular, its rapid rise to the forefront of the global consciousness means it is subject to the same scrutiny as other major social platforms.