Close Menu
    Europa

    Die EU wird Unternehmen ab Sonntag dazu verpflichten, KI-generierte Inhalte zu kennzeichnen

    Share.
    View 19 Comments

    19 Kommentare

    3. emotional_program0 on

      I’m all for this, but how will they realistically enforce it? What are the consequences for the companies that won’t do it?

    5. sandyWB on

      Fantastic news! 

      There are laws against false advertising so it made no sense to allow misleading AI slop.

    7. piggledy on

      This mainly applies to what they define as „Deepfakes“, i.e. AI imagery of real people, real objects, real locations etc. Looks like the main goal is to discourage false information rather than slop.

      Anything that has the chance to mislead needs to be labelled, but it looks like imagery that doesn’t look obviously real (illustrations, 3D images, paintings etc) doesn’t need to be labelled per se.

      It says „artistic, creative, satirical, fictional“ are exempt.

    8. No_Aesthetic on

      EU doing everything it can to limit the ability of domestic companies to compete at all. Why is the EU so dedicated to snuffing out every new tech-related industry?

    10. Far-Reaction-1980 on

      Aldi already labels Ai content
      and yes they show some of their products with an ai generated background

    11. Over-Willingness-933 on

      Most of the time you tell what is AI especially on YouTube. It’s really bad and often low quality with repetitive and boring quality. The warning will help to avoid it

    12. ButWhatIfPotato on

      Somehow youtube decided that the only thing I wanna watch is AI channels with less than 100 subscribers. Had to use plugins just so I can only see videos from my subscriptions so I can make the site bearable again.

    13. Euphoric-Witness-824 on

      This is awful! But thankfully our senators in the USA will do anything and everything to defend AI corporations against the evil tax paying citizens! I just saw an hour long and very passionate speech from Mitch McConnell about his love of AI that he delivered yesterday and then he did an amazing tap dance number! Praise our glorious politicians! For without their love we may have to live with slightly lower Dow Jones and healthcare for everyone … even the gross poor people and browns! In GOP Jesus name. A men. 

    14. TheDonnARK on

      And why the hell not?  The companies rolling out machine learning content want the revenue and attention given to actual people, it’s just that instead of one creator working at it and trying to find a niche, the companies roll out 100 slightly variated „creators“ and even if a few only get slight attention, they profit.  How quickly that attention (and this, revenue) would drain away if people know it wasn’t an actual person..

    17. EmbarrassedHelp on

      > That includes adding visible labels as well as technical markers, such as watermarks or embedded metadata, that help identify synthetic material.

      So called „invisible watermarks“ damage the content (like damaging the pixels of an image and forcing certain language choices in text), and would only serve to poison smaller open source AI efforts seeking to train with synthetic data.

      Metadata, hash databases, and other methods would be a safer bet for this. Companies should be forced to share what they are doing to the AI content you view, including watermarking, metadata, and other tracking attempts. I also worry we may see these requirements expanded to track non-AI types of content, which would be dangerous for privacy.

      > The requirements apply mainly to content created in professional contexts. Material designed to inform the public about matters of general interest must carry a label if it is produced by AI without human editorial oversight. At the same time, the EU has carved out exceptions. Individuals using AI for personal purposes are not covered, and exemptions exist for „artistic, creative, satirical, fictional“ works.

      The issue here is that these companies not likely to allow their users to take advantage of any exemptions. Its not financially in their interest to do so as it undermines selling their own AI products, while large corporate buyers can afford to pay for such exemptions. EU citizens will be stuck relying on China and foreign run tools outside the reach of EU law, if they wish to use synthetic data.

    19. Nazamroth on

      I can only hope this will also apply to „youtubers“ and their deluge of AI generated historical slop. It is hard enough to discern truth from myth as it is. Now we have this shitstorm muddying history for a few thousand farmed views to deal with on top of that.

    Leave A Reply