Our weekly Briefing is free, but you should upgrade to access all of our reporting, resources, and a monthly workshop.
This week on Indicator
We published an investigation from Keenan Chen into the phenomenon of YouTube videos that falsely claim to show police racially profiling and arresting a person that turns out to to b an FBI agent. The videos he identified have earned over 300 million views. In response, YouTube banned 13 channels.
Sympathy scams have proliferated across platforms this year, especially on TikTok. Craig revealed that the latest (and most concerning) iteration uses fabricated racial slurs and insults to bait people into buying cheap products. TikTok removed roughly 200 accounts that we flagged.
A reminder that our October workshop is today at 1 pm ET. Craig will give a hands on look at how to use digital ad libraries in investigations. We’ll look at real world examples and will share tips for getting the most out of ad libraries. Our workshops are for paid members. Upgrade now and we’ll send you the link to join, as well as the recording and transcript.
Deception in the News
Grokipedia launched on Monday as a self-styled alternative to Wikipedia. While some pages are simply an AI-generated regurgitation of the very Wikipedia it’s supposed to replace, other entries reflect the biases of its main owner, Elon Musk. The entrepreneur’s page, for instance, “describes him in rapturous terms while downplaying, or even omitting, several of his controversies,” according to Time.
Topics like climate change, George Floyd, and vaccines have an editorial slant in line with the billionaire’s views. The entry on Russia’s invasion of Ukraine takes a both-sides approach to the war and sourced some information directly from the Kremlin’s website. The page for the January 6 attack on the US Capitol goes out of its way to legitimize Trump’s false claims of election fraud and downplays the violent actions of rioters. The list goes on.
On Grokipedia, Musk’s allies are praised: the German AfD is defended from accusations of being a far-right party; Italy’s Prime Minister Giorgia Meloni is singled out as “distinguishing her[self] from predecessors in a political landscape often critiqued for elite detachment from empirical voter concerns.” This highly opinionated take is sourced to Meloni’s bio page on the Italian government’s website, which says nothing of the sort.
As with anything AI-generated, there are silly mistakes. Platformer’s Casey Newton, for one, has a recurrent bit about a non-existent wife. His Grokipedia page took that at face value and claims he is “married to a lawyer.”
Musk has responded to criticism of Grokipedia as expected, characterizing it as self-serving and biased.
At 404 Media, Jason Koebler wrote that:
“I have no doubt that Grokipedia will fail, like other attempts to ‘compete’ with Wikipedia or build an ‘alternative’ to Wikipedia, the likes of which no one has heard of because the attempts were all so laughable and poorly participated in that they died almost immediately. Grokipedia isn’t really a competitor at all, because it is everything that Wikipedia is not: It is not an encyclopedia, it is not transparent, it is not human, it is not a nonprofit, it is not collaborative or crowdsourced, in fact, it is not really edited at all.”
I tend to agree. The question, however, is whether Grokipedia can succeed not as an alternative to Wikipedia but as a means to delegitimize it. “Flooding the zone with shit” as a strategy doesn’t necessarily seek to replace institutions of collective knowledge generation; it seeks to deface them. — Alexios
📍 AI-generated misinformation and out-of-context media about a chaotic and lethal police raid in Rio de Janeiro reached millions on social media, according to Aos Fatos.
📍 NewsGuard spotted that an AI-powered Community Note contributor inaccurately claimed that MSNBC ran footage from an older protest and said it came from the recent “No Kings” event in Boston. Despite failing to reach consensus (to the credit of Community Notes’ bridging algorithm), the screenshotted note was used by several people to accuse MSNBC of lying. Senator Ted Cruz was among those who posted — and later deleted — the false claim.
📍 On YouTube, more people were watching an AI-generated crypto-hawking livestream of Nvidia CEO Jensen Huang than the real company event. The scammy stream was the top search result on YouTube for relevant queries. The channel that hosted the stream used a grey emoji in its name to try and mimic YouTube’s verified badge. It’s has since been deleted.
📍 The Washington Post found that the Trump administration has used “misleading footage in at least six videos promoting its immigration agenda” on social media. The footage is “muddying the reality of events in viral clips that have been viewed millions of times.”
📍 404 Media reported that top venture capital firm Andreessen Horowitz invested $1 million in a startup that operates a phone farm and sells social media manipulation services.
Tools & Tips
📍 SPOTTR is a tool that allows you to upload a video and search within the footage for specific objects.
I did a search for “cyclist” in the demo video that SPOTTR makes available. The tool highlighted several points in the video when one or more cyclist was shown in the frame. You get 3,000 credits when you sign up and an additional 1,000 credits per month on the free plan. (via Cyber Detective)
📍 Sofia Santos is well known in the community for her fun and challenging OSINT Exercises. She recently posted a reminder that you should take time to use and practice a new tool or technique, as opposed to stockpiling items without integrating them into your workflow:
Stop adding random content to your brain. There's no need to keep reading blogs, watching videos, and bookmarking OSINT tools like squirrels preparing for the winter months.
If you don't apply that knowledge, practice that technique, or explore that tool, chances are you will either forget about it, or it will be rendered useless.
So every so often, stop and apply what you learned.
📍 WebRecon is a new web reconnaissance tool built with Python that scans a site for email addresses, social media accounts, HTML comments, PDF files, and much more.
📍 The latest edition of Alicja Pawlowska’s OSINT newsletter includes a ton of great info about investigating the Russian internet, courtesy of Vytenis Benetis of i-intelligence.
📍 The OSINT Newsletter launched a new bi-weekly series of free posts aimed at teaching beginner-level OSINT skills. The first edition is “Username OSINT That Goes Beyond Tools.”
📍 If you’re not already using the great InVID Chrome extension to assist with image and video verification, among other things, Valentin Porcellini gave a nice overview of some of its features in a presentation at the Disinfo 2025 conference.
📍 OSINT Industries published a new guide, “Google OSINT: Using Google Profiles in OSINT Investigations.”
Reports & Research
AI Rihanna in a musical propaganda video endorsing Burkinabè dictator Ibrahim Traoré
📍 The European Council on Foreign Relations published a deep dive into information operations that support Russian-aligned regimes in Africa. One of the breakout stars is Ibrahim Traoré, who became president of Burkina Faso after a military coup. Among other findings, the report notes that there’s an AI-generated YouTube video of Eminem and Rihanna singing Traoré’s praises. It earned 2.3 million views and was monetized. The video was removed after Indicator reached out.
📍 Researchers from the Institute of Strategic Dialogue (ISD) found that “OpenAI’s ChatGPT, Google’s Gemini, DeepSeek, and xAI’s Grok are pushing Russian state propaganda from sanctioned entities.”
📍 Italian fact-checkers Facta found that athletes are frequent targets of deceptive, scammy, and hateful AI-generated content.
📍 Open Measures’ research team published, “Rumble’s Power Users Promote Crypto Scams and AI Slop.” It analyzed a year’s worth of Rumble videos and their associated comments and found that the most active accounts “flooded the platform with harmful material, and many exhibited unusual or suspicious posting patterns,” and that “likely scammers were targeting both Rumble content creators and audiences on the platform.”
Want more studies on digital deception? Paid subscribers get access to our Academic Library with 55 categorized and summarized studies:
One More Thing
Earlier this year, Meta CEO Mark Zuckerberg talked about how AI will automate many aspects of digital ad campaigns:
“We want to get to a world where any business will be able to just tell us what objective they’re trying to achieve, like selling something or getting a new customer, how much they’re willing to pay for each result, and connect their bank account and then we just do the rest for them.”
So how’s that going? Have a look at an AI-optimized ad that Meta auto-created for True Classic, a men’s clothing brand:
Indicator is a reader-funded publication.
Please upgrade to access to all of our content, including our how-to guides and Academic Library, and to our live monthly workshops.
