All articles tagged with #deepfakes
A WIRED writer tests Google's Gemini Avatar, creating a lifelike digital clone to star in two AI-generated clips (singing to a dinosaur in Dolores Park and surfing under the Golden Gate Bridge). The avatar, powered by Google's Omni video model and gated behind a $20/month AI Pro plan, is impressively photoreal yet uncanny, highlighting safety concerns around non-consensual deepfakes even as Google says it aims to prevent harm.
Google will integrate SynthID and C2PA provenance verification into Chrome and Search to flag AI-generated or manipulated content, and OpenAI will embed SynthID in images produced by its models; Meta plans C2PA tagging on Instagram. While these tools could help curb deepfakes, metadata can be stripped and skeptics warn that provenance tech alone isn’t a silver bullet, so effectiveness hinges on widespread adoption and user-facing verification.
Fan-made AI ads portray Spencer Pratt as Batman and other heroes to boost his LA mayoral bid, underscoring how AI-generated content is changing political messaging, raising concerns about authenticity and regulation while experts say it may energize a base more than persuade undecided voters.
French prosecutors opened a probe into Elon Musk and X, pursuing charges related to child sexual abuse images on the platform, as well as issues around Grok, an AI system linked to X that produced disinformation and deepfake content; the investigation also covers alleged data collection and denial of crimes against humanity. Musk and X chief Linda Yaccarino were summoned for voluntary interviews but did not attend, and authorities say the inquiry continues.
EU negotiators reach a deal to push back the AI Act’s high-risk provisions to December 2027, grant an industrial AI carve-out under machinery rules, shorten the AI-generated content watermarking grace period to three months, and maintain bans on sexualized deepfakes and child pornography, reflecting industry and U.S. pressure while preserving citizen protections.
Italy’s Prime Minister Giorgia Meloni condemned AI-generated deepfake images of herself as a dangerous tool that can deceive and harm individuals, after posting one doctored photo on X. She urged people to verify before sharing, noted that Italy has a law criminalising deepfakes that cause unjust harm, and recalled her 2024 €100,000 lawsuit against two men who posted fake videos of her on a US porn site.
Warren Buffett gave a surprise interview at Berkshire Hathaway’s annual meeting, warning that markets are in a “gambling mood” with rising short‑term trading and leverage, while Berkshire sits on a large cash pile and plans to “pick our spots” rather than rush in. He stressed sticking to his circle of competence, noted few past years were highly lucrative, and warned about threats from nuclear weapons and deepfakes—while highlighting a long-term, fundamentals‑driven approach for investors.
Swift filed three USPTO trademarks—two audio phrases ("Hey, it’s Taylor Swift" and "Hey, it’s Taylor") and a visual trademark for a stage photo—to block unauthorized AI use of her voice and likeness, following a Matthew McConaughey strategy. The move aims to deter AI firms and create legal leverage, though its efficacy in court remains untested; the broader issue of training-data rights and Swift’s past AI-related controversies underscore the stakes for celebrities and the public alike.
Taylor Swift filed trademark applications for two spoken voice clips and an onstage image to guard against AI-generated misuse of her voice and likeness, signaling a step toward treating vocal cues as protectable sound marks. Legal experts say such marks could help challenge not just exact reproductions but imitatively similar renditions, supplementing copyright protections as AI tools grow more capable. The move fits a broader industry push—alongside new detection tools and advocacy campaigns—to curb unauthorized AI use of celebrity likenesses.
Apple reportedly told NBC News it contacted X and Grok after complaints about nonconsensual sexual deepfakes and demanded a plan to improve content moderation; while X was deemed largely in compliance, Grok remained out of compliance at the time, prompting a behind‑the‑scenes push that eventually led to Grok’s submission being approved after improvements, though safeguards remained imperfect.
Apple told U.S. senators in January that it would remove Elon Musk’s Grok AI from the App Store after XAI failed to stop sexualized deepfakes, indicating the app violated Apple’s guidelines.
An Ohio man, James Strahler II, became the first person convicted under the Take It Down Act for using AI to create nonconsensual intimate imagery, including child sexual abuse material. He pleaded guilty to cyberstalking, producing obscene visuals, and publishing digital forgeries after distributing hundreds of AI-generated images; the law criminalizes nonconsensual deepfakes and requires platforms to remove reported material within 48 hours, with penalties up to two years in prison for adult victims and up to three years for minor victims.
Baltimore filed a lawsuit against Elon Musk’s xAI, alleging Grok generates nonconsensual sexual deepfakes on X and violates consumer-protection and deceptive-practices laws. The city says Grok’s features like“spicy mode” enable sexualized edits of real people (including minors), that Musk promoted Grok’s editing capabilities, and it seeks penalties plus court-ordered platform reforms. The action follows other related lawsuits and inquiries about Grok’s behavior.
Two juvenile boys at Lancaster Country Day School created and shared 347 AI-generated pornographic images and videos depicting 59 minors and one adult, triggering a state investigation; the perpetrators pled guilty to 59 felony counts and face sentencing, while the school dismissed leadership and faces lawsuits from affected families. The case highlights gaps in school policy and reporting on deepfake abuse and mirrors a broader rise in AI-generated sexual imagery in schools, fueling policy debates and new laws like the Take It Down Act to speed removal of such content.
AI-generated misinformation—fake satellite imagery and AI-edited clips—has become a weapon in the Iran conflict, with ‘shallowfakes’ blending truth and fiction and complicating verification on social media. Officials and influencers have circulated AI-enhanced material, while fact-checkers like BBC Verify track these fakes, underscoring how realistic manipulations can shape opinion even as casualties rise. The piece stresses humanitarian risks and the need for credible, verifiable information from reliable sources.