Become a fan of Slashdot on Facebook

 


Forgot your password?
Close
typodupeerror
×
China AI

China Wants Red Flags on All AI-generated Content Posted Online 58

Posted by msmash from the how-about-that dept.
China's internet regulator has proposed a strict regime that will, if adopted, require digital platforms to label content created by AI. From a report: The Cyberspace Administration of China announced its draft plan, which will require platforms and online service providers to label all AI-generated material with a visible logo and with metadata embedded in relevant files. The draft proposes that logos appear in several locations in a text, image, video, or audio file. In audio files, Beijing wants a voice prompt to inform listeners about AI-generated content at the start and end of a file -- and, as appropriate, mid-file too. Software that plays audio files will also need to inform netizens when they tune in to AI content.

Video players can get away with just posting notices about the content at the start, end, and relevant moments during a clip. Netizens who post AI-generated content will be required to label it as such. If they use generation tools provided by a platform, they'll be required to identify themselves -- and a log of their activities will be retained for six months. Some labels denoting AI-made content will be applied dynamically, based on metadata embedded in AI-generated content.
This discussion has been archived. No new comments can be posted.

China Wants Red Flags on All AI-generated Content Posted Online More

China Wants Red Flags on All AI-generated Content Posted Online

Comments Filter:

  • Good (Score:5, Insightful)

    by gweihir ( 88907 ) on Tuesday September 17, 2024 @09:52AM (#64792743)

    Now, why is China here the one that does the right thing and the oh-so-righteous west lags behind?

    • Comment removed based on user account deletion

    • China attempts to control all information. There can be benefits to this sort of thing, but overall we don't buy into it.

      • Especially if they want to use AI generated images without the flagging so you'll assume it can't be fake.

        • Expect that 99.9995% of all internet content will be AI generated. By sheer volume, any sentient analysis of them will fail to find useful information.

          It will make the internet near unusable as the chance of finding accurate data or factual content will be so low and drive internet users away.

          Same as when a well known musician release dozens of cover albums, at some point no one will care or listen.

    • Figures you'd want trigger warnings on AI images rather than simply Freedom. Seems consistent with your authoritarianism and anti-free speech stance.
      • I donâ(TM)t understand if someone posted a doctored video of you saying how much you love naze germany with added naze symbols tattooed on you, you want that up and around?
        • In a world were such videos are common, the credibility of the video creator is severely impacted. Folks are going to have to learn that not every video is real. If we can teach three-year-olds this lesson about television & cartoons, we can certainly expect adults to clue in, too.

    • Re: (Score:3)

      by Qwertie ( 797303 )

      As an anti-misinformation measure I fully support this and thank China for leading the way. But do not expect that the CCP will follow this law when it comes to their own information operations.

      It's actually more likely that if western governments had AI watermarking laws, they would follow them, since they know (or, at least, are slowly learning) that disinformation doesn't pay once it is uncovered. China can simply censor Chinese who discover and spread news of a CCP disinfo operation (or, as a last re

      • Re: (Score:2)

        by AmiMoJo ( 196126 )

        It would be extremely unusual if any government applied these kinds of laws to itself. Even in Europe, if you look at things like the ECHR, they basically say that "national security" overrides most things.

      • China did not "disappear people" since decades.
        They simply make law, and convict them under that law and put them into jail. That is much more simple.
        No idea what kind of content the CPC, could generate with AI, and then fail to watermark it?
        What scenario do you have in mind?

    • Because without China being the enemy, a whole slew of business and politicians would be losing out on income.

    • The Chinese regulators are right in concept, but ultimately it will be a feckless exercise. Legitimate sources will comply and sources with malicious intent will not. It will not make unflagged images more trustworthy, though I don't think that is China's ultimate goal.

      The cynic in me says this is part of the broader Chinese effort to promulgate their regulatory concept on to the broader Internet and not a noble desire to reduce the harm that AI modified images can cause

    • Now, why is China here the one that does the right thing and the oh-so-righteous west lags behind?

      How is this the right thing? What qualifies as AI generated? What happens if I use AI for edge detection in a fill routine? Is that AI generated art? If so, literally everything will have to have the label which will render it useless because many creative tool flows have some sort of AI in them to assist.

  • I for one, would welcome this (Score:3)

    by Savage-Rabbit ( 308260 ) on Tuesday September 17, 2024 @09:56AM (#64792753)

    I would welcome this, if only so that I could filter out all the shitty AI generated art currently flooding the internet.

    • Re: (Score:2)

      by gweihir ( 88907 )

      Exactly.

    • Are you gonna need trigger warnings and flags for Photoshopped images, too?

      • Are you gonna need trigger warnings and flags for Photoshopped images, too?

        No, Photoshopped images are nowhere near as low quality, irritating, or pervasive as the uniformly awful AI generated garbage.

        • Re: (Score:2)

          by suutar ( 1860506 )

          But they are similarly fictitious. If you just want high-quality bullshit, I feel like just waiting for the quality to improve is going to be more effective than instituting flagging.

          • and it will get better too. the 'quality' arguments are already losing water, but there's still something there for now. Eventually, and I mean like sometime next year maybe, the 'quality' argument will be dead.
    • Maybe AI can filter out the shitty AI generated art? It can be an AI death match

      • Maybe AI can filter out the shitty AI generated art? It can be an AI death match

        But what if AI starts filtering out Adobe's stock photo gallery since these days it mostly consists of AI generate crap art ... oh, wait, never mind, that's not a bug it's a feature.

    • Like video thumbnails that show some group by the hundreds doing some really bad thing, and the video creator down samples the images to hide the weird alien looking text on signs and people with two left hands. And without looking at the video and just scrolling through the list, people think it:s a real photograph of a real incident.
    • First of all, it doesn't work, because it's click-a-button trivial to remove any watermarking, (Yes, even the stegonographic kinds that people can't see.) So the label will be misleading, by default. What the label will eventually come to mean is "I don't have the expertise to remove this."

  • Evil bit returns (Score:3)

    by xack ( 5304745 ) on Tuesday September 17, 2024 @09:58AM (#64792759)
    Now with AI. Don't forget the DNT flag too.

  • We need cryptographically secure identification of AI generated images and text. They can be cool and fun, but should be labelled accurately

    • until you crop them or resize them, or upload them to facebook where they get re-compressed. destroying any signature. you could check the original and see if it's similar in some registry or something, but youd still be looking at an unverified copy most of the time.

  • "China's internet regulator has proposed a strict regime..."

    If it was the US government proposing this, would it be described as a "strict regime"? What does this mean exactly?

  • Because ... China?

    • Little red flags that say Made in China on all AI content. Because it's inevitable that China is going to be the AI leader.

      • Re: (Score:2)

        by AmiMoJo ( 196126 )

        As well as developing the hardware needed for AI domestically, China is way ahead of us on renewable energy deployment. They are already advertising services as greener because they can power them with wind and sunshine instead of gas.

  • Red Flags (Score:5, Funny)

    by Surak_Prime ( 160061 ) on Tuesday September 17, 2024 @10:25AM (#64792821)

    Aren't *most* flags in China red?

    Yes, yes. I'll see myself out.

    • Aren't *most* flags in China red?

      Yes, yes. I'll see myself out.

      Yeah, so red flagged content is good in China? Meaning red flagged AI content is approved by the government?

    • Good point. Using my babelfishing skills, it seems that "red flag" translates to Chinese as "danger signal".

      • I do not know about your country, but I in every country I have ever been: a ref flag is a danger signal, or s stop signal.

    • Re: (Score:2)

      by Bob_Who ( 926234 )

      Aren't *most* flags in China red?

      Yes, yes. I'll see myself out.

      That makes you Yellow , and now we can source the Supremacy of Orange aka Oompa-Loompa Luminosity.

  • In China, aren 't "red flags" a *good* thing?

  • Red China is going all in on being Red!

  • Me Chinese, me play joke Me put peepee in your Coke.

  • I actually think this is a good idea.
    I mean just from playing around with AI you soon discover that it often very authoritatively states something that is actually completely false.

    • you soon discover that it often very authoritatively states something that is actually completely false.

      ...as do politicians and random assholes on the internets.

      If you fail at critical thinking, what difference does it make?

      LLM != AI. The problem is the media and... welll.. humanity in general. But that is not a new problem.

  • Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press; or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances.

    -- First Amendment to the Constitution of the United States

    • It supposed to be a china flag emoji aka
      Unicode Code Points U+1F1E8
      U+1F1F3
      HTML Entities
      CSS \01f1e8\01f1f3
      JavaScript (JSON) & Java \ud83c\udde8\ud83c\uddf3
      PHP & Ruby \u{1f1e8}\u{1f1f3}
      Perl \x{1f1e8}\x{1f1f3}

  • I like the idea but I'd probably do it a bit differently. I'd focus on the AI software instead of the users. It should be impossible for end-users to accidentally fail compliance. The software should watermark by default, and failure shouldn't result in punishment of end-users, but a requirement for the company and/or social media platform to pull down the offending content. Of course there should also be a way to turn off the watermarks, but only by jumping through some hoops. Film makers need to be a

Slashdot Top Deals

If all else fails, lower your standards.

Close