Sensity

Generative AI crypto *scam* live now on YouTube Elon Musk appears to advertise a SpaceX crypto investment scheme during his Starship 5 speech Voice clone + high quality lipsync in loop. 50000 live viewers

Fake "live" videos on YouTube are the new frontier of deepfake scams

Sensity on The New York Times. A few weeks ago I had several conversations with Stuart Thompson, this article summarizes a fantastic investigative work that lasted for months. Since the release of our report, #thestateofdeepfakes2024, which highlights the rise of deepfake-powered scams on social media, numerous organizations have reached out to better understand the severity of this issue. We emphasized that these scams are mainly targeting vulnerable individuals within online communities, exploiting their trust and leading to devastating consequences. Thanks to the intelligence we provided, Stuart conducted an investigation into the impact of deepfake scams on society. He discovered that a Canadian man lost about $690,000 in a trading scam that began with an AI-powered video of Elon Musk asking for an initial investment of a couple of hundred dollars. Many other people worldwide are being scammed daily using the same framework. We've found that cybercrime groups from India, Eastern Europe, Russia, and China are leveraging these techniques globally. Once a scam performs well in one country, they immediately replicate it in others using text-to-audio, voice cloning, and lip-sync technologies. That’s the difference between Sensity and the competition: they comment on incidents after the fact on social media, while we are on the frontline mitigating the threat in real-time. Stay tuned; big announcements are coming in the next months…

I'm pleased to see European news outlets also referencing our latest report, #thestateofdeepfakes2024. Our threat intelligence approach has provided evidence that AI-Powered scams are widespread on major social media platforms, with thousands of incidents reported in the USA, Europe, and across Asia. Personally, I have always believed that #deepfakes targeting victims' money are far more insidious than those simply aiming to influence people for electoral or geopolitical purposes. In the report, we analyze the anatomy of these scams: how they are conceived, prepared, and launched, leveraging the commodification of tools for lip syncing and voice cloning, which have now achieved impressive quality at very low costs. Thanks to Oceane Duboust

The State of Deepfakes 2024 is available on https://lnkd.in/dP_hbr9 In the past 12 months, we have witnessed a massive deployment of #deepfake tools for creating and spreading influence campaigns, online scams, and KYC identity frauds. Thanks to our multilayered approach, which combines AI-powered detection forensic technologies with our unique deepfake threat intelligence system, we were able to collect a vast amount of material. This has allowed us to deeply understand the threat landscape, classify bad actors, and model strategies to enable companies, government agencies, and NGOs to fully grasp where the real threats lie and how to prepare for future challenges. #thestateofdeepfakes2024 #scams #influencecampaigns #KYCfrauds

I am thrilled and honored to be here in Latvia at the Riga Stratcom Dialogue, organized by the NATO Strategic Communications Centre of Excellence. This is an occasion to showcase Sensity's All-In-One Deepfake Detection solution to all government agencies of the NATO countries participating.

Amazing investigative work by CNN has revealed how Russian and Chinese propaganda are exploiting artificial intelligence to spread #deepfake influence campaigns on social media. We assisted the investigative team in confirming that these videos were AI-generated, demonstrating once again our reliability in developing cutting-edge deepfake detection tools for complex analyses. https://lnkd.in/eAWcTjXy

Pleased to have been at Visa #visasecuritysummit last week in Dubai to discuss the latest threats from generative AI and deepfakes to KYC and payment. Thanks Charles Lobo and the whole team at Visa!

How well are deepfakes being combatted on both a technical and legislative level? Fascinating start to the week attending INTERPOL's Synthetic Media Webinar this morning, hosted by Christopher Church. Pavel Goldman-Kalaydin, Sumsub, with insight into the deepfake creation process and the advances in recent years, showing what deepfake detection is up against with faceswaps, neural enhancers, and generative documentation templates. Natália Fritzen, also Sumsub, covered international legislation; from China, who seem to be taking the most stringent action against deepfake vendors and for users, to the EU and USA, who present a more indirect, open to interpretation, and arguably weaker protective regulation front against deepfakes for citizens. Of course, all eyes are on deepfakes with it being election year for so many countries around the world. Politicians are taking it more seriously because it affects their chances of office and may mean potential ruin. Tech companies have the spotlight on them as a result. Trust and Safety teams see the importance of countering this phenomenon and are taking steps. Boardrooms definitely won't want the bad press of letting deepfakes run rampant during election year and will have to be seen to be delivering in the fight against. But what happens next year? Will politicians and tech companies take their foot off the pedal as the issue becomes less of a threat for a while, until the next election cycle? Will Trust and Safety teams see a continuance of layoffs that started last year? What then for ordinary people affected by deepfakes? As Louise Hooper pointed out to me - "There is an interesting failure to consider existing criminal law and to what extent this covers synthetic content". That currently, there is more of a view to treating deepfakes as a data/civil issue centred around fraud than focus on its darker use as violence against women. Lastly, Giorgio Patrini, of Sensity, with a cool dive into just how vulnerable commercial vendors are to deepfakes when it comes to ID verification, as their penetration testing proved (9/10 were bypassed). Also, his technical slides on audio generated AI and use of spectogram comparison between real and fake speech was just cool. With so much interest in age assurance technology, I wonder how vulnerable these vendors are to deepfakes now and how vulnerable they'll be in the future as the tech improves and detection fights to keep up. What will that mean for the safety of users? Young users in particular? As the speakers today suggested, it will take a comprehensive approach that combines AI fighting AI, clever exploration of current legislation, and more specific legislation in future. #deepfake #ai #trustandsafety #onlinesafety

Generative AI and deepfakes are more and more often used to bypass traditional KYC checks. This is the end of KYC as we know it today. In 2022, we found that 10 of the most popular biometric KYC vendors were severely vulnerable to realtime deepfake attacks. (And so may be your bank, insurance or health providers.) The situation hasn't changed in 2024. KYC must be updated to counter new GenAI-powered attacks. Read more at: https://lnkd.in/dP_hbr9 Sensity on TechCrunch: https://lnkd.in/gKZDHAvd