[The PC Magazine report below describes a recent deepfake video of a keynote address that was streamed as the actual address was being delivered by the CEO of NVIDIA. The incident is just one example of how AI-based videos have expanded to deceive viewers of real-time events. On his blog, technology writer Shelly Palmer had this take:

“The good news is that no credible reports suggest the scam moved NVIDIA’s stock price or resulted in major financial losses. The bad news is that this kind of impersonation is not only super-easy to pull off, it’s getting harder to detect. Get ready for a flood of high-quality counterfeit events at scale. Livestreams and public appearances are new attack vectors; corp comms, IR events, and even internal town halls can be faked in real time. Detection systems are improving, but they still lag behind the speed and quality of synthetic media.

It’s time for countermeasures. Verify every event channel, watermark official streams, and do what must be done to monitor platforms in real time. Explaining why 100,000 people just watched someone pretending to be your boss is not a great way to start your day.”

An extended excerpt from a related story from Mediaite about AI-generated videos that fooled a well-known news organization also follows below. –Matthew]

[Image: Credit: X/DylanOnChips]

Did You Fall for This Deepfake Live Stream of Nvidia’s CEO?

As Jensen Huang gave his GTC DC keynote, a deepfake version was live streaming and pushing a dangerous crypto scam. It attracted many more viewers than the legit presentation.

By Jon Martindale
October 29, 2025

If you thought something was off during Nvidia’s October 28 GTC keynote, you might have been watching a deepfake stream. As the real event unfolded in Washington, DC, an AI version of the same event was also live streaming and attracted almost seven times the viewers.

Unfortunately, deepfake CEO Jensen Huang wasn’t discussing the future of GPUs and AI data centers, but was instead promoting a cryptocurrency scam with an appropriately dangerous QR code that you should definitely never scan.

CRN senior editor Dylan Martin spotted the fake stream and tweeted about it before it was taken down. Unfortunately, however, the stream was live for nearly an hour on a channel called NVIDIA Live. It was also the top result on YouTube if you searched “Nvidia GTC DC” at the time, according to TechRadar. This gave the scam a potentially huge reach, with the stream peaking at 95,000 viewers. While some of them were probably bots, that’s a lot more than the 12,000 who were viewing the official stream at the same time.

Martin got a transcription of deepfake Huang’s presentation, and it showed the phony Huang talking up Nvidia hardware’s capabilities for cryptocurrency mining. The stream then posted a QR code that would help viewers get set up with a new “crypto distribution” platform. [See Martin’s post on X]

Those who saw the stream claim it wasn’t hugely convincing thanks to stilted delivery and odd pacing. For a casual viewer, however, especially someone unfamiliar with Huang’s presentation style, this might be easy enough to miss.

Ironically, this comes one week after YouTube launched a new Likeness Detection tool, designed to help online platforms combat deepfakes. Did Nvidia not have access to the tool?

In a post-Sora 2 world, these scams are likely to proliferate. Deepfakes are much easier to create, so it’s essential that we all learn how to spot them and always verify that the video is legitimate and originates from a credible source before trusting anything it says or shows us.

Scammers pulled off a similar crypto scheme last year with an AI-generated Elon Musk.

—

[From Mediaite]

‘Gobsmacking’: Media Industry Observers Call Out Fox News’s Website for Getting Duped By AI-Generated Video, Then Attempting to Sweep it Under the Rug

By Joe DePaolo
November 2, 2025

Fox News’s website is drawing heat from media industry observers for getting duped by an AI-generated video — and then failing to take sufficient accountability for the error.

In a piece published Friday to Foxnews.com, writer Alba Cuebas-Fantauzzi posted about SNAP beneficiaries threatening to loot stores amid the government shutdown. The piece, originally headlined “SNAP Beneficiaries Threaten to Ransack Stores Over Government Shutdown,” quoted Black women claiming to be SNAP recipients complaining about the cutoff of benefits due to the shutdown.

“I have seven different baby daddies and none of ‘em no good for me,” one purported SNAP recipient said.

The site was promptly called out for running with a phony video. In a 7-minute video titled “Fox News Got Duped by AI—and Lied About It,” The Bulwark’s Tim Miller trashed the site for “horrific news judgment” in falling for the fake footage.

“Fox News got fooled by a series of racist AI videos purporting to be Black women complaining about losing their SNAP benefits, and they reported those ‘complaints’ made computers pretending to be humans as if they were real news,” Miller said. “Yikes.”

After getting called out, the headline of the post was amended to read, “AI videos of SNAP beneficiaries complaining about cuts go viral.” The change was accompanied by a brief editor’s note at the bottom.

“This article previously reported on some videos that appear to have been generated by AI without noting that,” the note read. “This has been corrected.”

[snip to end]