Our weekly Briefing is free, but you should upgrade to access all of our reporting, resources, and a monthly workshop.
This week on Indicator
Jane Lytvynenko, an award-winning digital investigative reporter, wrote a guide to using Telegram in digital investigations. It covers built-in search functions, downloading full channel histories, finding related channels, extracting file metadata, which third-party tools are worth your time, and more.
Craig investigated “lawslop” — a relatively new and thriving YouTube niche of AI slop that’s focused on courts, legal issues, and law enforcement. He found 25 channels that racked up over 1.5 billion views and often failed to properly label the content as AI-generated. Some channels mixed real footage with AI narration, or falsely said that synthetic content was taken "from credible, publicly available sources."
Deception in the News
India's Prime Minister Narendra Modi takes a group photo with AI leaders including OpenAI CEO Sam Altman (C) and Anthropic CEO Dario Amodei (R) at the AI Impact Summit in New Delhi on February 19, 2026. (Photo by Ludovic MARIN / AFP via Getty Images)
📍 Stricter rules on “synthetically-generated information” go into effect today in India. The regulations introduce new obligations for platforms to detect and label deepfakes and requires them to remove “unlawful synthetically generated information” within three hours of being notified. The human rights nonprofit WITNESS warned that “the framework relies on enforcement mechanisms that are technically unfit for purpose, while missing the opportunity to build on effective standards infrastructure.” It proposed “four urgent changes” to improve the rules.
📍 During a parliamentary inquiry on Monday, Australian Senator Michelle Ananda-Rajah, deputy chair of the Select Committee on Information Integrity on Climate Change and Energy, expressed serious skepticism about Meta’s plan to roll out Community Notes in the country.
📍 “Collaborative notes” written by AI are starting to show up underneath posts on X. You can check out the latest stats on Indicator’s dedicated tracker.
📍 Four years after the full-scale invasion of Ukraine, Russian state-controlled media are still able to get around EU sanctions and show up in Google Search results. Correctiv found more than 20 mirror domains for RT[.]de alone.
📍 Open Measures identified several 4chan threads where users exchanged tips on how to generate deepfake nudes of female Olympians, and sometimes shared their results.
📍 In an op-ed for The Guardian, British Prime Minister Keir Starmer announced that his government will require tech companies to take down nonconsensual deepfake nudes within 48 hours of being notified. (This is the same turnaround expected by the US’ Take It Down Act.)
📍 The WSJ reports that “Israel has arrested several people, including army reservists, for allegedly using classified information to place bets on Israeli military operations on Polymarket.”
Tools & Tips
Blacklight is a free tool that analyzes the tracking technologies used by a website. It can tell you about advertising, analytics, and other trackers that are present, and check if the site captures keystrokes or is recording sessions. It also checks if the Meta/Facebook pixel is present on a site. I often demo Blacklight during workshops because it’s an easy tool to check a site’s ad partners and third-party trackers.
The Markup, the nonprofit newsroom that makes Blacklight, recently announced that it now scans for the presence of the TikTok and X pixels. If you visit a site that has a Meta, X, or TikTok pixel, the companies can gather data about you and potentially link it to your activity on their platform. As The Markup notes:
When its pixel is embedded across many websites, the platform can compile a user’s data to build a detailed profile of their interests, behavior and other personal information. These profiles allow other businesses to buy ads from the platform to target categories of users — though this data can also be used for other purposes.
If you want to learn more, The Markup has a great series about the Meta/Facebook pixel. The BBC also recently published a story about the TikTok pixel.
In terms of using Blacklight, it’s very simple. Just enter the URL of the site you want to scan. You can also customize things a bit by clicking on “Options” in the search bar, which brings up this menu:
It allows you to set the location and device for the visit/scan, You can also force Blacklight to run a new scan, as opposed to pulling data from its cache. — Craig
📍 Full Fact and Maldita are the new operators of MediaVault, a project that originated at the Duke Reporters’ Lab to archive images and videos that have been analyzed by fact-checkers. It’s a useful tool (and associated database) that does a decent job of archiving social media. Full Fact noted that MediaVault “remains free to access for fact-checkers, journalists, technologists, and researchers who register on the platform.”
📍 Andrea (Drego) Draghetti made updates to Tosint, “a Python tool to analyze a Telegram bot token and a target chat ID (group/channel) using Telegram Bot API.” It extracts metadata that can be helpful in an investigation.
📍 Dorkwright is a “Python-based Google dorking utility that automates the extraction of file links from Google search results and optionally downloads them.” It uses Playwright to run searches and extract the results. (via The OSINT Newsletter)
📍 Flowsint is “an open-source OSINT graph exploration tool designed for ethical investigation, transparency, and verification.” (via Anastasios Vasileiadis)
📍 Sam Piranty of the BBC wrote a fascinating case study that reveals how a sofa and bricks shown in a bedroom photograph provided essential clues for catching a child abuser.
📍 David Clarke wrote, “Why multilingual Open-source intelligence (OSINT) matters more than ever.”
📍 Benjamin Strick published the latest edition of his Field Notes newsletter. It featured a wealth of tools and guidance for tracking ships, using satellite imagery, and more. Strick called it “a special edition for those seeking images from above.”
He walked through a few case studies and concluded with a point about the importance of combining tools/techniques:
You don’t need the most expensive tool or the most classified source. You need the patience to look at a problem from three or four angles and stitch the evidence together until it tells a story someone can’t wave away. That’s true whether you’re tracing sanctions evasion or documenting what happened to an airfield in Khartoum.
Events & Learning
📍 Registration is open for the 2026 edition of GlobalFact, the premiere conference for fact-checkers. It will be held June 17-19 in Vilnius, Lithuania.
Reports & Research
📍 A new study published in Nature analyzed the effect of X’s algorithmic feed on more than 1,000 US-based users during a 7-week period in 2023. The researchers conclude that “[p]articipants who switched from the chronological to the algorithmic feed prioritized a conservative policy agenda over a liberal one by 0.11 s.d., compared with those who remained on the chronological feed.”
📍 Forbidden Stories reviewed a trove of 76 leaked documents that revealed information about Russian influence operations across the Global South between January and November 2024. The activity was orchestrated by an entity referred to as “The Company.” It was run at least in part by former associates of Yevgeny Prigozhin, the deceased former patron of Wagner and the Internet Research Agency. The Company appears to have spent at least $7.3 million over ten months to promote false claims, such as that Ukraine supports terrorists in African countries.
📍 Renee DiResta writes that a prominent left-wing pundit’s conversion to 9/11 trutherism is “a tight case study in how conspiracy thinking travels now: not as a coherent alternative account of a situation, but as rhetoric that turns distrust and nihilism into the ‘reasonable’ stance.”
📍 Conspirador Norteño found that an obscure Bluesky account’s “starter packs” (lists of accounts to follow, typically organized by theme) appear to be driving large numbers of scammy and inauthentic followers.
Want more studies on digital deception? Paid subscribers get access to our Academic Library with 55 categorized and summarized studies:
One More Thing
Just a quick, disconcerting reminder that AI image generators are getting better all the time. Tom Jarvis of the Centre for Information Resilience shared an example of a synthetic image produced by Google Gemini that convincingly recreated buildings and other elements from a real reference image:
“We can see that the features on Independence Square in Kyiv are quite accurately reproduced, except for the sign behind the DJ,” he wrote. “The Pioneer DJ CDJs are correctly plugged into power, and the mixer is connected with RCA cables; even the faces are relatively convincing. Some minor details beyond the sign do stand out, but we are now firmly at the stage where geolocation alone is insufficient for verification.”
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.
