back to article Facebook won't nuke deepfakes? OK, let's tear up those precious legal protections from user-posted content, then

US lawmakers have warned they may revisit American tech corporations' blanket legal protections – specifically, the ones that shield internet giants from the fallout of user-posted content – in order to tackle the rise of deepfakes. A House Intelligence Committee hearing on Thursday dug into the issue of digitally altered …

  1. BebopWeBop

    The can of worms is open. So Facebook stick to their “principles” and leave Zuxk up (shame they paid less attention to Pelosi). The next one will push it a wee bit further - no “Gays orWhites or Jews or Blacks or Muslims or...” do X in the commentary, but getting closer. The next closer. And the next...

    Facebook will end up with a need to formalise their censorship, it will be expensive, law suits will emerge (left and right) and the inevitable will have become so.

    1. jmch Silver badge

      Google, FB, etc etc can't have it both ways. Either they have enough awareness and control of what content each of their users see (their marketing spiel to sell ads), or they don't have enough resources and control to promptly remove undesirable content (their 'defense' against regulation). They can't both be true.

      1. MonkeyCee

        Google ads

        I'm not sure about Google's vaunted targeted ads. Based on the ads it shows me and the product questions I get asked* it thinks I'm a lady.

        Bear in mind Google get paid more for ads to ladies, so it's in their interest to make me as valuable as possible.

        So I think they don't really have a clue what's going on either.

        * Which sanitary towel brand do you purchase? Uh... the same one as the empty packet I've been given by the lady I'm buying for....

        1. Argh

          Re: Google ads

          Being asked questions? Is this in Opinion Rewards? If so, one of the first questions it asks you is if you're male or female. A lot of people put female, as they get more cashback for it. Perhaps you did that and forgot?

  2. I.Geller Bronze badge

    AI database? It has causal relationships as chains of interdependent texts with their timestamps. And then a simple matching of a suspicious text (and all its interdependent texts and timestamps) with one that is trustworthy instantly reveals the suspicious' trustworthiness.

    1. I.Geller Bronze badge

      ‘Nothing,’ said Alice.

      ‘What do you know about this business?’ the King said to Alice.

      ‘Nothing,’ said Alice.

      ‘Nothing whatever?’ persisted the King.

      ‘Nothing whatever,’ said Alice.

      The intersections of the dictionary definitions for the words "business" and "nothing" with all dictionary definitions for other words (before/ after these) must be found, that is the same patterns must be detected. These intersections form AI database, this is Topology.

      1. I.Geller Bronze badge

        Encyclopedia for Alice's ‘Nothing’.

        12. The computer readable medium of claim 10, wherein the dictionary is selected from the group consisting of a locally stored dictionary, a remotely-stored dictionary, a locally-stored encyclopedia, and a remotely-stored encyclopedia.

        You can annotate words by an encyclopedia articles; where encyclopedia is a book or set of books giving information on many subjects or on many aspects of one subject and typically arranged alphabetically.

        For example you can annotate Alice's "nothing" by a definition from

        Or you may decide to use Encyclopedia Britannica, which in any case is billions of times cheaper than texts found somewhere.

    2. Throatwarbler Mangrove Silver badge

      Nice try . . .

      . . . but you're no amanfrommars1.

      1. I.Geller Bronze badge

        Re: Nice try . . .

        Trying to say it's all inter-connected. If you want to fight with fake news - you have to structure them into timestamp-marked patterns, and trace their origin and evolution.

        1. Charles 9

          Re: Nice try . . .

          Anything man can MAKE, man can FAKE.

          1. I.Geller Bronze badge

            I used to be a fake too.

            Yep, took five years for people here to read my words and understand me. I used to be a fake too.

            1. Charles 9

              Re: I used to be a fake too.

              And now you're a fake fake, right?

  3. Anonymous Coward
    Anonymous Coward

    Such efforts could undermine the public's faith in democratic institutions and in the media

    Errr no - they've done that all by themselves.

    As have politicians.

    They play the public as fools.

    1. I.Geller Bronze badge

      Blockchain technology in AI database as a means of combating fake news

      The US PTO is the closest in structure to AI database: everything is interconnected, everything has a unique timestamp: each patent is explained by all the others.

      The same goes for AI database: everything is explained by all the information and everything has a unique timestamp. If something is not explained - then clearly something is wrong.

      For example, if a patent came that is not referenced by other patents with unique time stamps - it is either a discovery or a forgery.

      Speaking of news, if the news does not have a pedigree with unique timestamps - it's either a sensation or a fake, which will be discovered at once. (2010, long before the word "blockchain" came)

      AI database is a blockchain database:

      14. The computer system of claim 9 in which said facility configured to extract predicative phrases is further configured to assign to the subtexts information regarding the date of their origin.

      1. I.Geller Bronze badge

        Blockchain technology is patented by the way. Guess who?

        By the way I got this blockchain technology patented long before the word "blockchain" became popular.

        1. I.Geller Bronze badge

          Lexical noise

          There is a sentence "Alice and Bob exercise merrily, she trains a lot."

          The word "merrily" can be an example of lexical noise; where it's typically superfluous patterns that do not explain the central themes contained within the digital textual information and, accordingly, removal of such noise often results in an improvement in the quality of the structured data.

          Suppose somebody doubts if Merrily the athlete who trains a lot being exercised by both Alice and Bob - it's a name. Or is it an adjective? Or both? So this sentence is structured having "merrily" both as a noun and adjective.

          With my patented lexical noise deletion AI-parsing gets FIVE patterns:

          - Alice exercise merrily - 0.25

          - Bob exercise merrily - 0.25

          - Alice trains a lot - 0.5

          - she trains a lot - 0.5

          - she trains merrily - 0.5

          AI sees that the word "merrily" is an adjective from its context and subtext, comparing its dictionary-encyclopedia definitions.

          Without AI-parsing gets SEVEN patterns:

          - Alice exercise merrily - 0.1(6)

          - Bob exercise merrily - 0.1(6)

          - merrily exercise merrily - 0.1(6)

          - merrily exercise a lot - 0.1(6)

          - Alice trains a lot - 0.5

          - she trains a lot - 0.5

          - she trains merrily - 0.5

          With dictionary-encyclopedia AI gets TWO synonymous clusters, without - THREE! One extra:

          - merrily exercise merrily - 0.1(6)

          - merrily exercise a lot - 0.1(6)

          If the search term "Merrily exercise a lot" came, the above sentence will be found even if it has nothing to do with Merrily!

          OpenAI and other companies which do not use my patented technology, which train their data by random data and not on dictionary-encyclopedia - cannot remove lexical noise, gets results like this with "Merrily" and are not trustworthy at all.

          Do not invest in them! You are gonna lose money!

          My AI database can, however, be 100% trusted.

  4. Wellyboot Silver badge

    US lawmaker decide to save own skin

    To stop the deep fake free-for-all from burning them, they'll happily throw Zuk & co. under a bus.



    The bus driver will be picked by lottery.

    Please form an orderly queue to purchase a ticket.

    1. iron Silver badge

      Re: US lawmaker decide to save own skin

      Can I buy more than one ticket?

      1. Anonymous Coward
        Anonymous Coward

        Re: US lawmaker decide to save own skin

        Can we use more than one bus?

  5. simonb_london

    Should apply to pubs also

    People in pubs can tell fake stories so convincing that the listener will have difficulty distinguishing fact from fiction.

    Therefore we should make the pub landlords liable for what the customers say on their premises.

    Oh, and people read bad things too, so we should make coffee houses liable for what books their customers read while they are there.

    1. jmch Silver badge

      Re: Should apply to pubs also

      "make the pub landlords liable for what the customers say on their premises"

      Well, people DO talk a lot of shite in pubs, but as soon as their business is affected, the culprits get thrown out pronto. The real issue is that many people go to the FB Pub specifically to consume shite, and advertisers don't care what shite is being posted as long as their ads appear.

      The key to draining the FB swamp is to somehow make it unprofitable for FB to show fake, illegal or otherwise undesirable content, but I'm buggered if I know how that would actually work.

      1. Charles 9

        Re: Should apply to pubs also

        Especially since (fake OR illegal) AND undesirable may well be a contradiction: forbidden fruit and all that. You could be staring at Prohibition all over again.

        1. Wellyboot Silver badge

          Re: Should apply to pubs also

          Prohibition won't be needed, Exemplary damages will be enough - $1m fine per view of any fake material.

          1. Charles 9

            Re: Should apply to pubs also

            Gotta be careful. Extreme fines have been challenged and I think even thrown out (or at least cut down) on Eighth Amendment grounds.

      2. simonb_london

        Re: Should apply to pubs also

        The key words here are "the culprits get thrown out" i.e. the pub itself isn't liable or whatever was said.

        Making online forums, of any description, liable for the behaviour of the individuals on it creates a dangerous precedent for them to start censoring speech just in case it lands the carrier in trouble.

        Some form of expression may be illegal but this is something that is to be determined /after/ it has been expressed, not preemptively before. Also the only crime is committed by the person that expressed it, not anyone who provided the space in which it was expressed. These are basic requirements of free speech being meaningful.

        1. jmch Silver badge

          Re: Should apply to pubs also

          "Making online forums, of any description, liable for the behaviour of the individuals on it creates a dangerous precedent for them to start censoring speech "

          And that's why it's very difficult to know where to draw the line on what is allowed and what is not, and, equally as importantly, WHO decides where that line is drawn. However there is an important difference between Facebook and any other online forum, and that is that FB personalises what each and every individual sees. If on a forum a comment i moderated, no-one sees it, and everyone else sees everything else. With FB, it chooses which of my friends' posts I get to see, and it does so in a way that maximises my engagement (and their profit). Although they don't edit any post individually, they edit my feed and therefore they are de facto exercising editorial judgement.

          The fact that they are delegating their moderation/approval process to an algorithm does not relieve them of the responsibility that goes with editing (rather than merely publishing) content. (Thought excercise - if the Times devised an algorithm to come up with stories, one of the stories coming up was both mistaken and extremely defamatory, and it got published on their website, you think they would be liable for that? You bet your arse they would )

          Online forums should not be liable for the behaviour/posts of the individuals posting to it, but FB certainly should

  6. Winkypop Silver badge

    OK I get it now

    This whole "Trump as POTUS" thing is just one giant deep fake exercise.

    I just knew it couldn't be real.

  7. Pascal Monett Silver badge

    "in order to tackle the rise of deepfakes"

    Right, because now that politicians have been the subject of such alterations, We Must React.

    That private citizens may have had their lives or reputations destroyed by such activity, or may risk that, is just collateral damage to the great march of Capitalism. We can't regulate that.

    But touch the politicians and ohh buddy, then you get their attention real quick.

    1. Wellyboot Silver badge

      Re: "in order to tackle the rise of deepfakes"

      You don't have to touch them (slimey), just wave a little cash, that'll work every time.

  8. imanidiot Silver badge

    The thing is

    When is a the news/history value of a "deep fake" video large enough that it should stay up? The very fact POTUS tweeted about it, and the discussion it caused would be reason to keep a video like the one of Pelosi up. At the very best it could include a short pre-roll warning it has been identified as a deep-fake.

    The second thing is that having a censoring mechanism like this creates a giant risk of actual censoring happening. If a REAL video is claimed to be fake by politicians, can they keep it out of the media attention simply by virtue of claiming it's a deep fake? It creates an opening that shouldn't exist imho. If the politicians want people to be able to judge real from fake and judge news for themselves, maybe they should adres the educational system and create actual critical citizens who can think for themselves. Of course no politician actually dares do that.

    Separate question, if a video actually IS fake but good enough to fool almost anyone, how are FaceGoogTube supposed to identify it as fake?

    1. jmch Silver badge

      Re: The thing is

      Problem is that polarisation has become so deep that it doesn't matter what's real or fake anymore. If you explain to a rabid republican that the video's a fake, they won't believe you. If Trump himself tells them it's a fake, they'll shrug and say "Yeah, THAT particular video is a fake but we anyway know Pelosi is a drunk (unpatriotic / white-hating / etc delete as applicable)". And the same if it's a fake Trump video to a rabid democrat etc etc.

      The real Trump undeniably (and even proudly) bragged about sexually assaulting women and was still voted for by (almost) half the US voting population. The 'drunk Pelosi' video may as well have been real, no democrat's ever going to change their vote over it.

      Truth doesn't matter any more, only perception.

      1. Bongwater

        Re: The thing is


        You have to stop spreading that fake crap.

        He said women let rich men do whatever they want to them.

        Want an example?

        OJ Simpson could still get dates, all that matters is that you are rich and famous. Women dated OJ AFTER he committed two murders. Rich men and women do whatever they want, that was his point and you fell for the trap of him being labelled misogynistic.

        Trump is an ahole sure but you only make these guys more successful when you repeat stuff that was never said, or taken into a context to get people to hate him more. You then identify yourself as a parrot, you offer nothing more but you keep repeating stuff that was never said.

        Nobody had a problem with the way Bill Clinton treated women, why is all of the sudden an issue with Trump?

        Congratulations to Toronto as well for winning the title!

        1. jmch Silver badge

          Re: The thing is

          @Bongwater - Your post is proving the point I made in mine. Thanks!

    2. jmch Silver badge

      Re: The thing is

      "if a video actually IS fake but good enough to fool almost anyone, how are FaceGoogTube supposed to identify it as fake?"

      I'm guessing that while the raw footage looks real, the underlying encoding might leak some clues as to fakeness that can be caught algorithmically. (of course that could be defeated by playing the fake on a high-quality monitor and recording the footage again on a camera). More succintly - employ an army of fact-checkers. Claiming it's too onerous is BS, their latest quarterly profits were almost $7bn. They could hire 1000s of moderators and barely tickle their bottom line.

      1. Charles 9

        Re: The thing is

        Wanna bet? And what's to say they can't make a DeepFake that can fool a HUMAN moderator? Besides, you're talking a small platoon versus a whole ARMY of clips uploaded every friggin' day!

  9. Andy The Hat Silver badge

    I don't quite understand why the people who post this material aren't implicated in any way.

    Whilst I don't like the Zuks etc of this world, if you are treating social media platforms as communication conduits rather than publishers, then the burden of blame for fake material falls onto the original posters.

    Personally I'd rather be in the situation where fake/copyright infringing/defamatory/politically sensitive stuff is able to be posted *but* the perpetrator of that post takes the full rap of the law for infringing copyright/defaming an individual/calling Trump names or whatever is illegal at the time. At the moment it appears that I can post defamatory material on the web and Google takes the rap for indexing it, not me for posting it. Whilst the whole Google monopoly is distasteful, that doesn't seem reasonable or correct.

    1. jmch Silver badge

      "if you are treating social media platforms as communication conduits rather than publishers"

      That's what FB et al WANT to be treated as, but FB is far from a 'dumb pipe'. It doesn't show every post from every one of my friends, it decides what to show, when and to whom. That's editorial control, whether there is a human making those decisions or not.

      "Personally I'd rather be in the situation where fake/copyright infringing/defamatory/politically sensitive stuff is able to be posted *but* the perpetrator of that post takes the full rap of the law for infringing copyright/defaming an individual/calling Trump names or whatever is illegal at the time."

      Problem being that the networks do not always know the real identity of the poster, and are rather quite happily to indulge the creation of fake accounts as that artificially boosts their membership figures.

  10. Anonymous Coward
    Anonymous Coward

    The problem is, it's a weapon ...

    and while you don't want your opponent(s) getting a-hold of it, you sure as hell ain't gonna give yours up either.

    Deepfakes are like manna from heaven for politicians ... they allow you to disown embarrassing past gaffes ("FAKE NEWS, FAKE NEWS !!!!") whilst simultaneously allowing you to have a pop at your rivals ("SEE WHAT THAT HILARY DID ?! THAT'S A TERRIBLE THING THAT IS.")

    I really can't see much being done.

  11. Eddy Ito

    To sum up

    Congress won't make a law abridging free speech they'll just make a law that punishes online entities who don't abridge free speech. What could go wrong?

    Will they punish SNL when Larry David or Alec Baldwin parody Bernie or Trump? Does it need to pass some nebulous "convincing enough" standard or is it some divine dowsing that that only Adam Schiff can adequately do? It's all just more congressional overreach and authoritarianism.

  12. TheSmokingArgus

    Yes Destroy All SMall Content Publishers

    Gee how convenient as they look to grant government new enforcement over speech on the internet, we see a story that looks to dissolve the Safe Harbors that protect bloggers from Copyright trolls.

    Gosh its almost as if the old media want to use government as a bludgeon to totally remove original small content creators & alternative views completely off the internet while cementing their control over new-media via the regulatory state.

  13. Henry Wertz 1 Gold badge

    Pelosi video

    Regarding the Pelosi video... the fact that ANYONE thought that showed her slurring her words really makes me despair for the lack of common sense of my fellow Americans. Like a SECOND into the clip I was like "why are they playing this clip 3/4 speed" . That's all it is, a clip played somewhat slowed down with pitch correction.

    I'm torn on what to do about this...the fine line is between getting rid of (or at least prominently labelling as fake) these manipulated items. But section 203 is also the legal protection for Google and the like which can't be realistically expected to personally review the billions of items they index.

    1. Anonymous Coward
      Anonymous Coward

      Re: Pelosi video

      There are those who argue that, as much money as they make, they need to either eat their profits or risk losing one of their main revenue streams to regulation.

      1. Anonymous Coward
        Anonymous Coward

        Re: Pelosi video

        Agreed. And all of the "evidence" is right there in a single spot - on Facebook's servers. It's not like these tech companies would need to somehow invade all of the homes across America to snoop on who's doing what. That would be creepy, crazy, and prohibitively expensive for any single company to do. No company could patrol that many citizens in real time.

        Sorry, hang on a sec...

        "Alexa play some BeeGees music"

        Why does she always veer off from soft rock to Linkin Park?

        Anyways, there is no way corporate America can keep a handle on what millions of citizens arw doing or posting. Only a desperate government could pull off that stunt.

  14. src

    Section 230 too permissive.

    Section 230 should be re-written so it forces platforms to allow speech permissible by the constitution. Social media companies then have a choice: editorialise all their content so they can be advertiser friendly or allow free speech and be safe from law suits. Advertisers will then have to choose between large, economically viable free speech platforms or small, expensive carefully monitored platforms.

    1. Anonymous Coward
      Anonymous Coward

      Re: Section 230 too permissive.

      Who is going to use small, expensive, carefully monitored platforms?

      The security aspects of a website for people with 7 digit incomes who like spending a lot of money on consumer goods would be something of a nightmare.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like