#deep-fakes — Public Fediverse posts

Russland nutzt Deepfakes in Desinformationskampagnen gegen Ukraine und Westen.
Eine Studie entlarvt die Methoden.
#Russland #Deepfakes

https://www.spiegel.de/ausland/russland-so-nutzt-moskau-deepfakes-im-kampf-gegen-die-ukraine-a-a9063feb-ef4e-4eb9-8a46-5fe016f78873?sara_ref=re-so-app-sh

Poliisi tarkentaa väärennettyihin tekoälykuviin Lappeenrannassa joutuineiden uhrien määrää – heitä on noin neljäkymmentä

Lappeenrantalaisen alaikäisen pojan epäillään luoneen tekoälyllä alastonkuvia useista alaikäisistä tytöistä.

https://yle.fi/a/74-20244668

#Rikosepäilyt #Deepfakes #Deepfake #Syväväärennökset #Lappeenranta #Eteläkarjala #Kotimaa

Some of the most widely shared content from the Iran conflict so far has been AI-animated LEGO videos released as part of a government's messaging. This article calls it "slopaganda," and both the U.S. and Iran have been using it from official accounts.

We have laws of war for autonomous weapons, while governments posting AI-generated memes from official channels have no rules at all, and the author makes a good case that this is a form of militarized AI too. Just a non-lethal one. For years, security awareness training told people one simple thing: when in doubt, go to the official source. That advice stops working when the official source is posting memes.

A few things this changes for anyone running security or communications:

- Your employees are being trained, every day, to expect official accounts to be casual and unreliable. That makes impersonation easier and weakens the "does this sound right?" instinct.

- AI-generated content is now cheap enough that a nation-state, a scammer, and a bored teenager can produce it all at the same quality. The content itself tells you nothing about who made it.

- Verification has to move from "does this look real" to "did this come through a channel I can confirm?" That means signed messages, known domains, and a phone call when it matters.

I'd love to see the same energy that goes into AI weapons policy applied to AI messaging from official accounts. In the meantime, keep your own company's official channels boring and factual. Boring is a security feature.

https://philosophynews.com/the-ethics-of-slopaganda/

#Cybersecurity #AI #Disinformation #security #privacy #cloud #infosec #DeepFakes

Das Schweizer Parlament beschäftigt sich aktuell mit mehreren Vorstössen, die die Erstellung und Verbreitung nicht-einvernehmlicher sexualisierter #Deepfakes unterbinden wollen. In unserem neuen Explainer erläutern wir Ursachen und Folgen von sexualisierter Gewalt im digitalen Raum. 👇👇
https://algorithmwatch.ch/de/sexualisierte-gewalt-im-digitalen-raum/

The Chain of Custody for Reality: Why Every Public Image, Voice, and Document Must Prove Its Origin

https://bolesblogs.com/2026/09/01/the-chain-of-custody-for-reality-why-every-public-image-voice-and-document-must-prove-its-origin/

Colorado Newsline: Colorado Sen. Cathy Kipp submits first complaint under state’s deepfake disclosure law. “Sen. Cathy Kipp, a Fort Collins Democrat, filed a complaint with the office on Aug. 5 against the Independence Institute, as first reported by CBS Colorado. She alleges that a video posted to multiple Independence Institute social media accounts violates a 2024 law that requires […]

#Steady #Klimacrew

#Deepfakes sind eine wachsende #Bedrohung, da sie täuschend echt wirkende #Fälschungen ermöglichen.

Können Menschen, auch bei Vorwarnung, #Deepfakes von echten Videos unterscheiden? #Wissenschaftler sind dieser Fragestellung nachgegangen, denn wenn #Fälschungen das Internet überfluten, kann das Vertrauen in digitale Inhalte und die #Medien insgesamt untergraben werden.

https://tino-eberl.de/missbrauch-kuenstlicher-intelligenz/deepfakes-warum-warnhinweise-nicht-ausreichen-um-faelschungen-zu-erkennen/

#Desinformation #FakeNews #KIMissbrauch #Retröt

Republicans Calling News AI Is Becoming the Right’s Favorite Excuse for Facts They Don’t Like

https://thedemocracyadvocate.com/trump-watch/republicans-calling-news-ai/

Jopa yli sadasta tytöstä on levitetty tekoälyllä tehtyjä alastonkuvia Lappeenrannassa

Tapauksesta on tehty poliisille useita kymmeniä rikosilmoituksia.

https://yle.fi/a/74-20242691

#Rikosepäilyt #Kaakkoissuomenpoliisilaitos #Syväväärennökset #Deepfakes #Lappeenranta #Eteläkarjala #Kotimaa

Italy passes groundbreaking #AILaw targeting criminals who make deepfakes: https://zorz.it/RENqB

#MattGrowcoot #EUAIAct #GenAI #law #cybersecurity #Italy #PrivacyProtections #criminals #deepfakes

Schon krass, was schon geht...

heise+ | IT-Sicherheit: Wie ein KI-Klon einen Kollegen per Videocall überlistet | heise online https://www.heise.de/hintergrund/Taeuschen-mit-KI-Ein-Deepfake-Experiment-im-Video-Test-11322994.html #heiseplus #ArtificialIntelligence #AI #DeepFake #DeepFakes #Digitalkompetenz #Bildung #education #Datenschutz #privacy

AI scams are becoming much harder to spot.

The FBI recorded:

• 22,364 complaints involving AI-related information in 2025
• More than $893 million in reported losses

AI can now help scammers clone voices, create convincing videos, generate fake profiles and write personalized messages that sound completely natural.

So do not rely only on what you see or hear.

https://thenewsink.com/ai-scams-are-becoming-harder-to-spot/

#AIScams #Deepfakes #Cybersecurity #OnlineFraud #ArtificialIntelligence #TheNewsInk

They Dedicated Their Lives to Teaching. Then the Deepfakes Started

https://fed.brid.gy/r/https://www.wired.com/story/teachers-deepfake-ai-students-content/

"So we have a mistargeted law, badly drafted to go after a theoretical problem that hasn’t proven to be real, combined with penalties harsh enough that companies like Anthropic comply in the broadest, bluntest way possible — and all of it dropped into a world that has increasingly decided that AI use is a binary good-vs-evil question, and where no nuance is allowed.

The watermark won’t actually catch the people it’s ostensibly aimed at — those seeking to deceive people. Those people will likely be sophisticated enough to remove any such watermark and walk away clean. However, it will likely catch the people who had every right to use the tool in the first place: the non-native speaker who ran their draft through Claude to sound more fluent, the assistive-tech user who needed help composing a sentence, or the writer who wants an extra level of review on any text they’ve written. They’re the ones who will wear the scarlet letter. The EU wanted to look tough on AI and Anthropic wanted to look compliant. But neither of them has to answer to the the disabled user who used the tech to help them communicate, who now gets accused of lying and cheating."

https://www.techdirt.com/2026/08/20/the-eu-wanted-a-deepfake-detector-it-got-an-ai-scarlet-letter/

#AI #EU #AIAct #GenerativeAI #Watermarking #Deepfakes #NannyState #Anthropic #Claude #LLMs

Scary. Regulate AI or shut it down.

In an interview with The Economist, Yuval Noah Harari says AI has become so good that he didn’t recognise a deepfake of himself for a minute and a half. Speaking to The Economist’s editor-in-chief, Zanny Minton Beddoes, the author and historian warns that this technology is coming for all of us.

#ai #artificialintelligence #deepfakes #theEconomist #democracy

https://www.youtube.com/shorts/eIlnq9__tF0

Habe gerade die Doku über Sir Michael Caine gesehen. Sehr beeindruckend, wie sich der Cockney-Junge zu einem der bekannstesten Schauspieler entwickelt hat. Und dass er sich gegen akustische Deepfakes engagiert, finde ich großartig.

https://t3n.de/news/michael-caine-deepfakes-1758926/?utm_source=t3n+Newsletter&utm_campaign=de159d409d-EMAIL_CAMPAIGN_2026_08_20_12_50&utm_medium=email&utm_term=0_faba600569-de159d409d-280234037&mc_cid=de159d409d&mc_eid=5480a8c3cc

#Deepfakes #MichaelCaine

The EU Wanted A Deepfake Detector. It Got An AI Scarlet Letter.

https://web.brid.gy/r/https://www.techdirt.com/2026/08/20/the-eu-wanted-a-deepfake-detector-it-got-an-ai-scarlet-letter/

56 studies. 86,000 people. One number tells you almost everything.

Humans catch deepfakes about as often as a coin flip catches heads.

https://shortlivedage.substack.com/p/you-cant-tell-a-deepfake-from-reality?r=8uodfo&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

#ai #artificialintelligence #technology #tech #deepfakes #deepfake #misinformation #practicaltech

Meta Ran Ads for an App That Promised to Nudify Female Politicians

https://fed.brid.gy/r/https://www.wired.com/story/meta-ran-ads-for-an-app-promising-to-nudify-female-politicians/

RE: https://mastodon.social/@404mediaco/117111454162065803

In this important interview of Hany Farid by @emanuelmaiberg, they discuss (among other concerns) the growing problem of synthetic #CSAM and other #NCII and the necessity of imposing much greater consequences on the executives and companies most responsible for creating and distributing these media. As Farid explains, perceptual hashing algorithms such as Photo DNA can't significantly hinder the promulgation of #deepfake media, because too many perceptually distinct #deepfakes can be produced too quickly.

I'd argue that the only real solution to the problem of synthetic abuse images would necessarily apply much earlier in the process: legally restrict the training data.

Make it a criminal offence, mandatorily piercing the corporate veil, extending from executives and board members to ICs, to produce, possess, distribute, wilfully retain, advertise, or make available any media synthesis model derived from data that includes or ever included both:

• any image, whether still or video, which depicts or appears to depict human nudity or human sex acts (regardless of whether any depicted person exists or ever did, and considering for this purpose human any fictional humanoid species portrayed as having visually similar intimate anatomy or capable of visually similar sex acts); and either or both of:
  • any image, whether still or video, which depicts or appears to depict any adult (as of the age at which they are or appear to be depicted) who exists or ever did, who did not, with unmitigated capacity to do so at the time, expressly provide free and informed consent not tied to anything else except financial compensation, to their image being included in a model capable of synthesizing intimate images, without ever withdrawing that consent before production of the model; or,
  • any image, whether still or video, of any minor (as of the age at which they are or appear to be depicted), whether ever existing or not.

Any model trained on both can produce NCII. To prevent synthesizing NCII, we must mandate the deletion and prohibit the production of models trained on both. Companies wanting to synthesize porn can bifurcate their training, so the only humans in their porn-capable model would be adults consenting specifically to that.

Allow, as a sole exception, solely for the purpose of preserving evidence of a crime, retention of encrypted copies of a violating model, made after cryptographically hashing the plaintext model and kept with the encrypted model, encrypted such that it is mathematically impossible to deduce or infer any part of the model without private keys kept in sole custody of a court of relevant jurisdiction or its designate. In all other cases and for all other purposes, impose an affirmative duty of immediate and irreversible deletion upon discovery (or, for pre-existing models, upon enactment of the law) of all violating models, by any entity with the technical and legal access to do so, and an affirmative duty to report to law enforcement the existence of such models by all persons with knowledge of their existence, as well as whether, to the reporting person's knowledge, the model was immediately and irreversibly deleted. Make failure to report existence of the model a crime for everyone who knows and fails to report it.

Retroactively and proactively forfeit as proceeds of crime all gross revenue or valuable consideration given or received by any and every entity (natural person or otherwise) wilfully involved in production, retention, making available, providing, accessing, or using such models, or for concealing any of the foregoing except by means of immediate and irreversible deletion, in consideration for any wilful part in the foregoing. Make all such revenue and the value of any such valuable consideration tax debts owing immediately, with compounding interest retroactively from the date of transfer of funds or consideration (or, for pre-existing models, from the enactment of the law). For the purpose of this forfeiture, value any such non-monetary valuable consideration equivalently to the highest monetary value it attains from date owing (as defined in the preceding) to date of full repayment.

Make it additionally and independently a civil offence, with a private right of action, retroactively and proactively exempted from all indemnity clauses, hold harmless provisions, mandatory arbitration clauses, class action waivers, limitations of liability, forum selection clauses, choice of law clauses, or any other terms of contract, licence, or agreement purporting to prevent or limit means of legal redress.

Make the only escape hatch from legal liability to destroy the model immediately if one can, or to report it immediately otherwise.

If Sam Altman, Sundar Pichai, Demis Hassabis, Alexandr Wang, Mark Zuckerberg, David Holz, Prem Akkaraju, Christobál Valenzuela, Anastasis Germanidis, Tim Flossbach, and everyone who works for their companies in any “AI”-related capacity—even an intern on their first day—could end up imprisoned and bankrupt for allowing a model whose training data makes it capable of producing NCII to continue to exist (or, for those who can't destroy the model, for failing to report it), the availability of NCII synthesis would diminish markedly.

Even though the current US regime of perpetrators of offences against children, bribed as they are to serve the interests of the media synthesis companies, would never enact any such a law, the CEOs and most of their company's employees do like to visit other countries, to fly over other countries' airspace, or to sail in other countries' waters. Meanwhile, several US states may be persuadable to enact such laws (albeit perhaps a fit filed down around the edges due to SCOTUS 1A interpretations), and preempting them federally would be politically self-poisonous. (What politician or candidate wants to out themselves as pro-NCII?)

Enact similar laws in enough jurisdictions, and in effect, the same might as well be enacted in the USA. Russia would undoubtedly be a hold-out, but for which: enact sanctions, and provide #Ukraine abundant means of kinetic enforcement of those sanctions.

#cdnpoli #eupol #ukpol #auspol #nzpol

This R-Rated Film Studio Wants to Be the HBO of AI

https://fed.brid.gy/r/https://www.wired.com/story/this-r-rated-film-studio-wants-to-be-the-hbo-of-ai/

AI-generated wildlife images are endangering animals and people, experts warn. https://www.japantimes.co.jp/environment/2026/08/17/wildlife/conservationists-ai-wildlife-visuals/?utm_medium=Social&utm_source=mastodon #environment #wildlife #orangutans #ai #wildlife #misinformation #deepfakes #malaysia

El lado del mal - La Policía Nacional detiene a un cibercriminal que usaba DeepFakes para obtener Certificados Digitales gracias a un Glitch https://www.elladodelmal.com/2026/08/la-policia-nacional-detiene-un.html #DeepFake #DeepFakes #IA #Cibercrimen #Identidad #KYC #DNI #Fraude #Ciberestafas

Operation Matryoshka: Russia using ‘BBC news’ fakes to pervert a key German vote.

With fake videos posing as BBC, ARD, or major German newspapers flooding X, TikTok, and Bluesky, millions of German voters risk making decisions in September based on AI-generated lies targeting Kremlin-critical candidates.

https://mediafaro.org/article/20260811-operation-matryoshka-russia-using-bbc-news-fakes-to-pervert-a-key-german-vote?mf_channel=mastodon&action=forward

#Germany #Deepfakes #AI #Disinformation #BBC #ARD #X #Bluesky #TikTok #SocialMedia

Facial recognition is the future of identity confirmation systems.

Facial recognition can be used to ascertain someone's age when someone tries to access online content deemed to be adult in some way.

It's a completely secure way of identifying people, so they tell us.

/s

https://www.theregister.com/security/2026/08/11/deepfake-hiccup-unmasks-suspected-digital-certificate-fraudster/5285934

#ageverification #facialrecognition #deepfakes #theregister #elreg