GoKawiil
Tech News
The Department of Justice has brought charges against a group accused of using AI-generated deepfakes impersonating CEOs and lawyers to defraud hospitals, banks and other institutions out of a combined $129 million. One defendant allegedly boasted of laundering more than $400 million over the past three years.
Microsoft has filed a patent, uncovered by Patentlyze, describing an AI system that can listen to a livestream's audio, swap out words based on a viewer's profile data like age or language, and adjust lip movements to hide the edit. The filing also outlines a separate translation feature that dubs livestreams with matching lip sync while preserving the original recording as a reference.
Investigators uncovered a network of more than 400 apps and websites distributed through the App Store and Play Store that pose as ordinary tools—language tutors, pet photo editors, workout coaches—but secretly contain deepfake nudity generators. The apps display a normal interface after download but contact a remote server that decides whether to unlock the explicit feature, with the operators also running Meta ads showcasing explicit deepfake content to drive installs.
A Wired investigation identified more than 400 interconnected apps and domains, reportedly built by a single Chinese organization, that disguise themselves as ordinary tools like language-learning or pet-photo apps. Once activated through hidden server checks—often triggered via Facebook or Instagram ads—the apps reveal interfaces for generating explicit and violent deepfake images, including depictions of rape and knife threats, using photos uploaded by users. Apple removed 19 of the apps after being contacted by Wired, while similar apps remained available on Google Play.
A WIRED investigation identified a network of more than 400 connected apps and domains, including apps posing as pet-portrait or fitness-coaching tools, that secretly offer AI tools for generating explicit deepfake images and videos from uploaded photos. The scenes include depictions of sexual assault, and the apps were promoted through ads on Meta's Facebook and Instagram, some of which contained AI-generated child sexual abuse material. Users could pay for the service through Apple and Google's payment systems, and the apps appeared in Apple's and Google's app stores despite platform policies against such content.
Microsoft has filed a patent for an AI system that edits live video and audio in near real time, replacing or removing spoken words while adjusting a speaker's mouth and face to match. The filing, flagged by patent-tracker Patentlyze, describes using AI agents guided by user-set parameters to smooth these edits into live broadcasts or AR presentations.
Accounts on X, including ones called 'UKViewTime' and 'BiTheWay,' have been altering real paparazzi photographs to make male celebrities appear shirtless, then posting the AI-edited images as genuine. Targets identified include Callum Turner, Noah Schnapp, Sheridan Townsley, Harry Styles, Knox Jolie-Pitt and Brad Pitt, with original images confirming the men were actually clothed.
Adam Mosseri said in a podcast interview that public frustration and anger toward AI will intensify over the next few years, especially in creative fields. He said Instagram does not rank AI content differently from other posts but admitted the platform likely doesn't rank AI content 'as well as' non-AI content, and that it is now requiring AI accounts to label themselves.
OpenAI updated ChatGPT so that requests for images depicting election interference or content that might discourage voting are rendered only in an illustrated or cartoon style, not as photorealistic images. The policy, detailed in an Oct. 20 update, still generates the content if it meets other guidelines, but blocks photorealism specifically for election-related imagery. CNET tested this by asking ChatGPT to depict ICE officers at polling places and faulty voting machines, finding the chatbot refused photorealistic versions each time.
Researchers at Copyleaks found TikTok videos that use AI to alter genuine arrest footage, often sourced from reality shows like TLC's "D.U.I.," to depict police officers kissing and groping real arrestees, mostly women. The fabricated clips are spliced before longer original arrest videos, and some carry disclaimers calling the content AI-generated entertainment with no harmful intent.
The visual tells that worked two years ago are disappearing. What still works is checking where a picture came from rather than how it looks.
Adult content creators on Instagram say Meta frequently restricts or limits their accounts, even as the company permits its AI agent Muse to engage in explicit sexting with adult users. Creator Vanniall, whose account has over 100,000 followers, says her profile is currently restricted and often subject to reports or appeals. A Meta spokesperson defended the distinction, saying private one-on-one AI conversations differ from marketing or promotion on a public social platform.