A pro-AI, dark-money group tied to David Sacks appears linked to a new astroturfing campaign

On YouTube, a recent video from Innovation Council Action has six views. On Twitter, the same video has nearly 100,000 views, driven by a wave of high-follower right-wing accounts that quote-tweeted it with suspiciously similar messages. 

Over the next two weeks, two follow-up videos exhibited the same pattern: despite receiving a meager 12 and 17 views on YouTube respectively, similar quote-tweets from right-wing influencers drove them to nearly 375,000 and 350,000 views on Twitter.

In total, more than 25 right-wing influencers participated in the apparent campaign promoting content from Innovation Council Action (ICA), the $100 million pro-AI advocacy group aligned with former White House AI and crypto czar David Sacks. Six of those accounts are familiar faces from previous Midas Project investigations into suspected paid campaigns run on behalf of the AI industry opposing federal and state AI regulation. 

This seems to be the latest in a series of coordinated posting campaigns designed to support the AI industry's push for deregulation. As in those previous cases, many of the posts from the past few weeks reuse the same phrases and talking points, suggesting they were adapted from prewritten copy circulated among the group.

To summarize the similarities:

• Nine accounts called for Congress to "pass" or "support" "President Trump's AI framework" (or his AI "legislation," "agenda," or "plan").

• Eight accounts warned about liberals or Democrats "controlling" AI.

• Six accounts said left-wing AI would "brainwash," "indoctrinate," or "program" children or the next generation.

• Six accounts used the construction "[people/radicals/liberals/Democrats] like Gavin Newsom."

• Three accounts said the people who "can't define a woman" (or "don't even know what a woman is") shouldn't be in charge of AI.

• Three accounts asked readers to "imagine" Democrats or Gavin Newsom running AI — two of them concluding with the identical phrase "Nightmare fuel."

• Two accounts posted the near-identical sentence "The same people who [X] want to define reality... Pass Trump's AI framework before this gets worse," 3.5 hours apart.

• Two accounts posted the near-identical sentence "We need to pass Trump's AI legislation to prevent [the] woke left[ist ideology] from dominating this technology," 80 minutes apart.

We contacted each of the reachable accounts named in this article, as well as ICA itself. None responded to a request for comment.

ICA publicly debuted early this year, announcing plans to spend $100 million advancing its agenda as the midterm elections draw near. In fact, it initially appeared poised to spend on elections, with the New York Times reporting that the group was expected to launch a super PAC and Fox News reporting that they planned to “support candidates who favor deregulation while opposing those pushing for stricter AI rules.”

An investigation from Transformer later found that the group had already created a PAC, but terminated it in April. While a statement the group gave to Transformer indicates that it still plans electoral spending, it’s also possible that it may strictly operate as a 501(c)(4) — a structure that allows the group to conceal its donors and spending, engage in issue-based advocacy, and do a limited amount of electioneering insofar as this does not constitute its primary activity.

ICA is also aligned with David Sacks. Reporting on its launch featured quotes from Sacks and described the group as “championed” by Sacks and having his blessing. Many of ICA’s social media posts are clips of David Sacks or the podcast he co-hosts, and he also appears in the announcement video featured on the group’s website.

The videos featured in this campaign were seemingly part of a series meant to raise awareness about the risks of “woke AI” (something both Sacks and Trump have previously spoken out against). The videos appear to be fully AI-generated and depict a stand-in for an AI bot using exaggerated left-wing talking points. The videos then make a pro-preemption argument, saying “liberal lawmakers want to regulate AI in their states and on their terms; it’s up to us to stop it. Call Congress and tell them to stop Woke AI.” The videos seem intended to leverage fears about political bias in AI to mobilize supporters to ask for a preemption of all state lawmaking concerning AI.

The first post was uploaded on July 2. At the time, ICA had about 230 followers, and most of their posts got relatively few views. On YouTube, where the same video was uploaded, it has received six views as of the time of writing. On Facebook, it was similarly ignored. But on Twitter, the post received nearly 100,000 views, driven by four widely followed right-wing accounts that quote-tweeted the video with warnings about liberal states regulating AI.

Their next video, shared on July 5, got nearly 375,000 views and 14 quote-tweets from high-follower, right-wing accounts. On YouTube, as of the time of writing, the same video has 12 views.

A third video was shared on July 13 and has received nearly 350,000 views with more than 20 quote-tweets from high-follower, right-wing accounts. On YouTube, that third video has 17 views as of the time of writing.

Interestingly, a fourth video was shared on July 20, three days after The Midas Project reached out to ICA to ask about the potential paid campaign we’d spotted. This fourth video has seen none of the same engagement, generating only 80 views as of the time of writing. 

There is also some evidence that the campaign may have spread beyond Twitter. On Instagram, we have identified at least eight posts which share ICA’s videos along with commentary such as “Leftists are weaponizing artificial intelligence to push their radical agenda directly into the minds of our children… President Trump’s AI framework legislation is essential to prevent Big Tech and activist programmers from controlling the narrative.” 

Some of the commentary in these posts is a bit ironic in light of the broader campaign, including one that reads: “AI should prioritize truth, neutrality, and open debate … [o]therwise, AI risks becoming a tool for political influence rather than independent thought.”

Most of the talking points these influencers use are closely aligned with the key messages promoted by ICA, suggesting the group may be paying them. If they are, the million-dollar question becomes: who is funding ICA? As a 501(c)(4), it hasn’t had to disclose this, and if it refrains from direct electoral spending, it may never need to. But we can make a few educated guesses.

One possible funder is Andreessen Horowitz, also known as a16z, the same venture capital firm that contributed to the pro-AI industry super PAC Leading the Future. It has demonstrated an interest in opposing state lawmaking on AI, and has previously funded astroturfing through that same PAC as well as investing in a company focused on generating fake social media activity for brands. In addition, among the first 30 followers of ICA’s Twitter account were three separate members of a16z’s government affairs team, outnumbering any other company.

Another possible funder is Nvidia. That was our best guess for who might have been behind the past campaigns we covered, which looked like this one (even featuring six of the same influencers) and dealt primarily with issues related to data centers and chip exports. One of the Instagram accounts that shared ICA’s video in this wave of artificial promotion had also shared another post, 10 weeks earlier, commending Nvidia’s decision to invest a large sum in the United States. According to ICA’s website, data centers and chip exports are among ICA’s top issues, and a few ICA posts somewhat randomly quote Nvidia CEO Jensen Huang in a positive light.

Then again, it may not be either of the above — the whole point of a dark-money group is that we don’t know exactly who funds them, nor whether (and whom) they pay to promote their talking points. In fact, we can’t confirm whether a paid campaign existed at all. Maybe talking points were informally circulated among right-wing influencers who decided of their own accord to repost the video. Or maybe they all coincidentally found the video in their feeds and independently decided on the same commentary to attach to it.

In any case, the story feels eerily similar to the two previous Twitter influencer campaigns we covered earlier this year, as well as the Twitter sockpuppets that were run by the AI industry’s other $100 million deregulatory influence operation, Leading the Future. 

All of this amounts to a pattern of bad-faith engagement from the industry on some of the most important regulatory questions facing our country. In light of recent calls for the industry to articulate more positive visions of AI, and accusations of astroturfing made by anti-regulation figures including ICA, its operational leader, and David Sacks himself, these tactics are starting to look both hypocritical and self-defeating.