Copyright pcgamer
Nvidia streamed the keynote speech of its latest GPU Technology Conference yesterday—though unfortunately thousands of folks seem to have fallen for a deepfake Jensen Huang extolling the virtues of cryptocurrency instead. Just when you thought the dodgy reuploaders and react channels were bad enough, it turns out you've now got to worry about fully counterfeit event streams too. For those that missed it, the official Nvidia GTC, originally streamed live from Washington, D.C. on October 28, can be viewed here. Though it's sitting pretty at nearly 200,000 views now, that wasn't the case during the stream; while about 20,000 tuned in for the official event as it went live, nearly 100,000 viewers were said to be watching the deepfake feed broadcast (hosted by a channel calling itself Nvidia Live) at the same time (via Tomshardware). The counterfeit stream has since been removed, but CRN editor Dylan Martin posted to X about the debacle as it happened, observing that at its peak, the official upload had only about a fifth of the viewership of the deepfake feed. The deepfaked Huang claimed to postpone the main event in favour of promoting "a crypto mass adoption event that ties directly into Nvidia's mission to accelerate human progress." The AI-generated Huang also encouraged viewers to scan an on-screen QR code, and send in supported cryptocurrencies. It's not yet clear how many (if any) viewers may have scanned this QR code or lost money while believing this fake stream to be legitimate. This turn of events is in many ways unsurprising. Just earlier this month, OpenAI CEO Sam Altman volunteered his own likeness into the maw of AI video generator Sora 2, with perhaps predictable results. As for Huang, this is the fourth GTC keynote he's hosted this year alone, likely adding to the wealth of easily accessible video footage for bad actors to use as AI training data. Please excuse me while I go to pick out a cute, brown paper bag to wear on my head for forever more. As Nvidia already secured a $4 trillion valuation earlier this year, (and is now fast approaching the $5 trillion mark), in no small part thanks to the many moves it's made around AI, there's an irony that isn't lost on me. It's also worth noting that, even as the money going into AI-based investments threatens to outstrip the money coming out, Huang recently downplayed concerns about a bubble. Nvidia NIM is already being leveraged in deepfake image detection models like Hive's, but I wouldn't be surprised if Nvidia itself made a move similar to Meta, and released more deepfake detection tools in the near future. Failing that, it seems we all need to keep our eyes peeled, as our online reality threatens to come apart at the seams.
