Our weekly Briefing is free, but you should upgrade to access all of our reporting, resources, and a monthly workshop. We’re offering a limited time 20% launch discount.
ICYMI: Indicator is the merger of Faked Up by Alexios Mantzarlis and Digital Investigations by Craig Silverman.
This week on Indicator
Alexios discovered that the single top contributor to Community Notes is a startup called Web3 Antivirus. What looked like a story about someone gaming X’s crowdsourced moderation turned into an example of the platform’s failure to combat crypto scams.
Craig did a deep dive into the ownership and image upload policies of 9 web-based facial recognition services. If you’ve used, or are thinking of using, a service like Pimeyes.com, FaceCheck.id, ProFaceFinder.com, Lenso.ai and others, our research has exclusive new information.
We also heard from a reader who said that the pornbait ads on Meta we wrote about last week are big in Bangladesh. The company removed another 475 ads we flagged.
Deception in the news
Notes into the void? X’s crowdsourced moderation platform was broken for several days, which meant that notes were invisible to users (h/t Maarten Schenk). The company acknowledged the problem with Community Notes on Monday, ascribing it to an overall platform outage. Alexios took a look at the data and saw a clear decline in activity. Ratings took a big hit, with only 19 notes reaching helpful status on May 26, down from an average of 130 the week prior.
The overall number of notes also fell significantly. But even at the lowest point there were still 1,200 notes being written a day (down from ~1,750). Maybe the program wasn’t down for everyone? Alternatively, users didn’t notice or didn’t care that it was and kept community noting into the void.
“Trust is everything.” We’re a little late to this, but Bluesky is expanding its verification process. Through a Google Form, “authentic and notable accounts” can apply for a blue check. Trusted verifiers — organizations that verify users — can apply through the same process. For example, NBC News is a trusted verifier and it ascribed a checkmark to friend of the newsletter Brandy Zadrozny.
AI deepfakes get supercharged. The Verge called Google’s generative AI video model “a slop monger’s dream.” Henk Van Ess used it generate 18 news videos for a completely fabricated local political scandal in less than 30 mins. Total cost? $8. “The barrier to entry for sophisticated political misinformation is now lower than the cost of a movie ticket,” he wrote.
AI roundup. Pope Leo XIV is already getting the disinformation treatment. American politicians are getting AI slop fan fiction. It’s possible that ChatGPT is being used to create receipts for an Italian North Face scam operation. Two fringe websites spent nearly €40,000 on Meta ads that spread AI-generated content and divisive messaging about last year’s DANA storm in Spain; whoever is behind the ads likely also ran an influence campaign during Poland’s presidential elections.
Other important things. Russian Telegram channels circulated a fake Euronews video targeting Moldova’s pro-EU leadership. Google’s Mandiant warned that cybercriminals are distributing malware by promoting fake AI video generator websites through social media ads. Civitai might finally be getting serious about fighting non-consensual deepfake nudes.
Sponsored: American politics has never been this chaotic.
Every week, independent writer and researcher Kyle Tharp shares unique data, original reporting, and insider analysis at the intersection of politics, media, and the internet via this must-read newsletter. Make sure you’re subscribed >>
Tools & tips
📍 doxcord is a “tool designed to scan Discord servers for social media links containing tracking parameters. The tool can identify Instagram, TikTok, and Facebook links that include tracking identifiers, and organize them by user and server.” (via OSINT_Tactical)
📍 OSINT Portal offers a useful collection of tools and resources. Worth bookmarking. (Via The Friday 5 Newsletter)
📍 Last week we highlighted the new YouTube Comment History tool from LolArchiver. 404 Media published a piece about its privacy implications, and LolArchiver responded.
📍 @cyb_detective created a Twitter list of companies that develop OSINT tools.
📍 CTRL-F: Find the Facts is a free online course that teaches news and AI literacy skills. Share it with friends and family.
📍 The Nieman Journalism Lab looked at “How this year’s Pulitzer awardees used AI in their reporting.”
📍 Mike Caufield offered helpful guidance on how to prompt LLMs in order to get better results for fact checking and research.
📍 We missed this when it was first published: the Global Investigative Journalism Network did a deep dive into Violation Tracker databases for the US, UK, and globally. They’re searchable databases of regulatory fines paid by companies. These are great tools to add to your company backgrounding process.
Events & learning
Global Fact 12 organizers announced that Brazilian Supreme Court justice (and Elon Musk nemesis) Alexandre de Moraes will speak at the fact-checkers summit in June. Register here.
Craig will be in Paris for the Viva Technology conference on June 13. His panel is, “Viral Voices, Virtual Walls: What is Social Media’s Role in Shaping Democracy?” Reach out if you’ll be there.
Reports & research
AI-enabled scams: Data & Society published, “Scam GPT: GenAI and the Automation of Fraud.” It warns that “the inherent ‘scamminess’ of the contemporary internet, and shifting economic norms blur the line between legitimate opportunities and scams, leaving even tech-savvy individuals vulnerable.”
Why not both? A group of practitioners and researchers reviewed the pros and cons of crowdsourced fact-checking, concluding that “community efforts, while valuable, cannot replace the indispensable role of professional fact-checkers.”
Unprepared. “Most schools are not currently addressing the risks of nudify apps with students,” concludes an important new report from the Cyber Policy Center at Stanford.
AI and elections. The International Panel for the Information Environment’s latest report found that generative AI content featured in at least 80% of elections held in 2024. Just this month a deepfake made a prominent appearance in the Buenos Aires elections.
Fake X accounts enter the trade war: Graphika uncovered “a network of over 1,000 X accounts engaged in coordinated and inauthentic behavior in an effort to influence the online information environment around the U.S.-China trade war and global tariffs.” The network’s operator(s) “were very likely aligned with the interests of China,” according to the report.
Want more studies on digital deception? Paid subscribers get access to our Academic Library with 52 categorized and summarized studies:
One thing to watch out for
Faking a run
Fake My Run is a new site that let’s you fabricate a running route for your app of choice. Draw your path on the map, choose a pace and date, and export the file. Developer Arthur Bouffard was inspired by Strava Mules — runners who get paid to juice people’s stats. He created the app to “prove the good old saying that you shouldn’t trust what you see on the internet” and “because it was technically doable and entertaining.”
Indicator is a reader-funded publication.
Please upgrade to take advantage of our limited time 20% launch discount. You get access to all of our content, including our how-to guides and Academic Library, and to our live monthly workshops.
