Society & Ethics

At Least 400 Fake Accounts Defended Nvidia's Jensen Huang

Midas Project researchers found at least 400 fake accounts defending Jensen Huang, linked to a PR firm. The network also lobbied on Nvidia's China chip stance for a year.

Fake Accounts Rushed to Jensen Huang’s Defense, a Trend That Should Worry Every Business
Fake Accounts Rushed to Jensen Huang’s Defense, a Trend That Should Worry Every BusinessBru-nO / pixabay
By Sophie Lindqvist7 min read

Why it matters

  • Midas Project found at least 400 fake accounts defending Nvidia CEO Jensen Huang, with links to a PR firm tied to earlier sockpuppet campaigns.
  • Huang resigned as president, CEO and board member on Nov. 12 after a board review opened Nov. 8, ending 33 years at Nvidia.
  • X has deactivated or suspended at least 260 of the accounts; the most-viewed post drew roughly 8,000 views and 32 likes.
  • The Midas Project traced the network to Koios, an Israeli firm that denies involvement; who paid remains unclear.
  • 88% of 232 Nvidia employees polled on Blind before the resignation said Huang should remain CEO.

At least 400 fake accounts rushed to defend Nvidia CEO Jensen Huang during his board investigation this month, according to research from the nonprofit Midas Project reported by The Washington Post. The accounts posed as ordinary people — a chef, a stock trader, a Hebrew teacher, several environmental activists — and had links to a public relations firm previously tied to sockpuppet campaigns.

The finding lands at a moment when Nvidia's leadership crisis kept the company in the headlines daily, and it shows how cheaply manufactured consensus can seep into the conversations that executives, journalists and lawmakers use to judge public sentiment.

Huang's real supporters included some of tech's most powerful names. Elon Musk, Sequoia Capital partner Shaun Maguire and Bill Ackman all publicly rallied behind him. The problem is that they were joined by a crowd that only looked grassroots.

How did the Nvidia crisis unfold?

Nvidia's board opened an independent review on Nov. 8. Four days later, the company announced that Huang had resigned as president, CEO and board member, effective immediately. The resignation ended a 33-year run at the company he co-founded.

Chief Financial Officer Colette Kress is serving as interim CEO. The board named Harvey C. Jones as Nvidia's first-ever chair.

That's the corporate timeline. The fake-account timeline runs alongside it, and it started long before the investigation made news.

Who ran the fake network?

Tyler Johnston, executive director of the Midas Project, published his findings on Nov. 10 — two days before Huang resigned. His research shows the network wasn't just defending Huang in a crisis.

Over the past year, the accounts made thousands of posts promoting Nvidia. They pushed the company's position on selling advanced AI chips to China and attacked lawmakers who support export controls. They frequently replied to journalists, former U.S. officials and members of Congress, including Sen. Elizabeth Warren, D-Mass., Republicans on the House Foreign Affairs Committee, and the House select committee focused on China.

The evidence of coordination, according to the Midas Project, is compelling:

  • The vast majority of accounts were created, renamed or given new profile pictures in the same sessions as other accounts in the network, often minutes apart.
  • Many accounts appeared to use scheduling software, with posts going live within the first half-second of a new minute.

The Midas Project traced the network to Koios, a small Israeli company that presents itself as an online financial fraud detection firm. Haaretz previously reported that Koios ran fake accounts in an earlier influence campaign. Koios claims to have gone out of business, yet the accounts kept posting. Koios has denied any connection. IntellACT, a firm that may be its successor, didn't respond to the Midas Project. Who paid for the campaign remains unclear.

Nvidia told the Midas Project it has "no relationship" with the companies linked to the accounts and is trying to identify the source.

The campaign's reach was limited. According to the Post, the most-viewed post drew roughly 8,000 views and 32 likes. Some accounts deleted posts after Johnston published his findings. X has deactivated or suspended at least 260 of the accounts.

Small reach doesn't mean small intent.

Why does a small sockpuppet campaign matter?

The goal of a sockpuppet campaign is rarely virality, according to the analysis. It's to manufacture the appearance of consensus in front of the people who matter. When a lawmaker, a reporter or an analyst checks the replies under a post and sees a steady stream of seemingly ordinary Americans voicing the same opinion, it suggests public sentiment leans a certain way — even when it doesn't.

Technology has made this cheaper than ever. Building a believable fake persona once took real time and effort. Today, generative AI can produce a realistic profile photo, a convincing bio and an endless supply of on-message replies in seconds.

In this case, the Midas Project concluded the accounts were likely run by people rather than bots, often with AI's help. One account slipped up and posted what appeared to be an AI prompt asking for a reply that would get engagement "while still sounding like a real human." The tools are cheap, widely available and improving daily, which means campaigns like this one will get harder to spot.

For businesses, the implications break into three concrete risks:

  • Sentiment is a business input. Boards, investors, communications teams and executives watch online reaction to gauge how a crisis is playing. With Nvidia's stock swinging on every new report, plenty of people watched the public response to Huang closely — and some of what they saw was manufactured. Decisions built on polluted data are bad decisions waiting to happen.
  • A company can be damaged by a campaign it didn't commission. Nvidia says it has no connection to the accounts, yet the network pushed the very positions the company has been lobbying for in Washington. That hands critics an easy way to cast doubt on Nvidia's legitimate advocacy and on every genuine supporter it has.
  • Real voices get drowned out or discredited. Once people learn a crowd was partly fake, they tend to discount the entire crowd. Genuine customers, shareholders and employees risk being lumped in with the bots.

What did real Nvidia employees say?

Inside the company, a very different kind of support for Huang emerged — one with names attached.

As WebProNews reported, some Nvidia employees pressed the board to bring Huang back. According to Wired, which reviewed internal correspondence, a longtime engineer submitted a post to an internal "suggestion box" before Huang resigned, urging the board to consider employee sentiment before taking action. The post has drawn more than 1,600 comments, several asking the board to find a way to reinstate Huang.

One employee went further, sending a companywide email that declared, "I stand by Jensen." Within hours of the resignation, roughly two dozen employees posted messages supporting Huang on LinkedIn. Wired said it could not find any employee posts backing his departure over the same period. In an anonymous poll on the workplace app Blind, taken Nov. 9 and 10 before Huang resigned, 88% of the 232 Nvidia employees who responded said he should remain CEO.

The contrast with the sockpuppets is stark. These are real people with names, badges, careers and, in many cases, significant equity on the line. They spoke through company channels or under their own names, and they will live with the consequences of the board's decision. One can debate whether they're right — the board evidently reached a different conclusion — but their voices reflect a real constituency. The sockpuppets borrowed the credibility of ordinary people without any of the accountability or the stake.

Even the internal side of the story isn't entirely immune to verification problems. A public website, StandWithJensen.com, hosts an open letter to the board from what it describes as self-identified current and former Nvidia employees, partners, shareholders and supporters. The site notes the letter is not endorsed by Nvidia or Huang. There's nothing to suggest the effort is anything but genuine. Still, self-identified signatures on a website are a far cry from verified internal correspondence — and that is precisely the point. Unless a voice can be tied to a real, accountable person, it's increasingly hard to know how much weight it deserves.

What's the takeaway for every business?

Huang's fate was decided by a board weighing an investigation, not by reply counts on X. But the episode should warn every company, investor and journalist who treats online sentiment as a reliable barometer of public opinion.

The lesson is simple, even if the practice isn't: treat online reaction as unverified until proven otherwise, and give voices tied to real, accountable people — employees, customers, shareholders — far more weight than an anonymous crowd. As fake personas become cheaper and more convincing, the ability to tell the difference may become one of the most valuable skills a business has.

Source: webpronews.com

Share this article:

More from Sophie Lindqvist

Sophie Lindqvist

Show full bio

Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.

749 articles

Related articles

  1. Nvidia’s Huang investigation and four stock-market precedents
  2. Tech leaders rally behind Jensen Huang after Nvidia CEO exits
  3. Nvidia board rules out Jensen Huang's return as reinstatement campaign passes 180,000 signatures
  4. NVIDIA Board Not Discussing Huang's Return, CNBC Reports
  5. Nvidia CEO Jensen Huang Steps Down, Sending Shares Down 10 Percent

« Previous articleNext article »