Zedstrian

The U.S. under Trump has shifted away from aiding Ukraine, so a graph that compares total aid since 2022 is somewhat misleading. It would also be useful to have an entry that combines the EU amount with those committed by individual EU nations as a better comparison to the figure cited for the U.S.

Just because information exists doesn’t automatically give AI companies the right to exploit it for use in their LLMs. People should still have the choice as to whether or not the data they produce is using as AI training data (and that choice should be a voluntary opt-in at the user level, not an automatic opt-in via automatic scraping or the sale of one’s data by a social media platform).

Many users left Reddit to avoid that happening to their submissions there, but if companies like Microsoft just then go and scrape the Fediverse, then we’re back in the same position.

AI companies scrape the web relentlessly in search of data to include in their models; searching for other instances, the extent of its knowledge demonstrates that Microsoft is non-consensually training its models on community posts, in addition to basic instance data.