In January, President Donald Trump shared two short clips from a popular YouTube account that claimed Walmart was closing hundreds of stores in California. The claims were false. California Governor Gavin Newsom’s communications team was alarmed not only by the falsehoods but by the unclear form and origins of the videos, two Newsom staffers told Semafor.
The @MeganWright62 channel in question largely hosted sensational AI-generated videos that purported to show companies abandoning California, grocery stores running out of food, and teachers quitting en masse. These videos were assembled by AI at scale with little regard for accuracy, yet they appeared polished enough that many viewers — or at least some YouTube commenters — did not recognize or did not care that AI had produced them.
Newsom’s team has spent recent months pushing for the removal of hundreds of such accounts as it increasingly focuses its media monitoring on AI-generated spam. Staff described campaigns pushing false claims such as Kroger closing a California distribution center, Costco and Shell shutting stores and gas stations, or GM and Ford closing California plants — the latter two companies having had no factories in the state for decades.
Accounts removed and the likeness protection program
Newsom’s office says it worked with YouTube to identify more than 700 accounts since January that were later removed for violating the platform’s terms of service via YouTube’s “likeness protection” program, which flags unauthorized uses of a person’s face in deepfakes. Newsom’s team said more than 100 of those accounts were populated with AI-generated content containing false or highly misleading information about Newsom and California’s purported decline.
Brandon Richards, a spokesman for the governor, told Semafor that “algorithms are rewarding lies and bad actors are taking advantage — particularly those who want to see a divided America.” Richards said major platforms like YouTube “have been a helpful and willing partner in this fight,” but warned that “the combination of bots and increased access to powerful AI, coming at a time with the doomscroll nature of our content consumption habits, has made it easier than ever to push rage-bait and false narratives that are believed on sight.”
Examples: channels, view counts and YouTube’s response
Many of these accounts attract little traction and go inactive quickly, but some — including @MeganWright, @PumpScore, and @TaxExodus — have amassed tens or hundreds of thousands of views. New York Reports, a channel that began posting earlier this year, initially used an AI-generated avatar to push a narrative of economic collapse tied to New York Mayor Zohran Mamdani and later posted error-filled videos with apparent real narration over repetitive AI-enhanced stock footage.
One New York Reports video, a mash-up of unrelated local news segments and articles, had nearly 300,000 views and about 1,300 comments when Semafor checked it. Another March video had 90,000 views; a separate AI-narrated, stock-footage-heavy piece reached 200,000 views.
After Semafor flagged New York Reports, Pump Scope, Tax Exodus, and similar accounts to YouTube, a company spokesperson confirmed they had been removed for violating the platform’s spam policies, which prohibit using automated tools or AI to churn out high volumes of repetitive content with minimal changes.
YouTube’s challenge and actions against AI 'slop'
YouTube itself has raced into AI, building new tools for YouTube Shorts and rolling out AI-generated summaries, transcripts, and interactive chat features. Still, the ease of producing cheap, low-quality video risks flooding the platform with garbage, a concern for executives trying to keep YouTube feeling premium. Neal Mohan, YouTube’s CEO, called combating AI slop a “top priority” in a public letter earlier this year.
In recent months the platform tightened rules around monetizing “inauthentic content,” including repetitive or template-based videos, AI-generated videos lacking a clear narrative, and content using AI personas to address sensitive topics. Rather than removing individual videos, YouTube has focused on accounts that post AI slop at scale to trick users and harvest cheap ad views.
Video-editing company Kapwing estimates that more than 20% of the videos YouTube’s algorithm shows to new users are low-quality AI-generated content, representing a growing share of overall views. Political content remains a relatively small slice of that total, but it can be especially volatile for platforms trying to avoid heated controversy.
Academic perspective and outlook
Columbia researcher Camille François told Semafor that understanding what political slop looks like on platforms and where it comes from is “a complicated and fascinating topic.” She pointed to a recent academic report on the emergence of low-quality AI-generated videos and tech platforms’ responses, noting that modest adjustments to ranking signals, monetization thresholds, or eligibility rules can change creator incentives without imposing a categorical ban on generative content.
YouTube is betting it can curb AI-generated spam by improving detection of unauthorized uses of public figures’ likenesses and training systems to automatically spot and label obvious AI-generated content. The company must monitor enormous volumes of content without letting too much garbage proliferate or unintentionally suppress legitimate speech — a balance that will be difficult to maintain.
Meanwhile, conservative media outlets have long produced bespoke, human-made content with similar aims, and the arrival of generative AI only complicates the landscape for political communication and platform governance.
Conclusion
YouTube and other major platforms recognize the risk that low-quality, AI-driven content poses, particularly when it targets political narratives and public figures’ images. Newsom’s team and other observers say the problem is currently manageable with platform cooperation, but without broader systemic changes it is likely to worsen.
