Briefing: A new tool for detecting AI images, the Macrons sue Candace Owens and more

Our weekly Briefing is free, but you should upgrade to access all of our reporting, resources, and a monthly workshop. Our next workshop, focused on website investigation tools and techniques, is on August 1. If you’re already a member, thank you for your support!

Craig reported exclusive findings from a new study about repurposed YouTube channels, and shared tips for how to identify potentially repurposed channels in the wild.

He also revealed that hundreds of Meta ads with fake virus alerts tricked people into downloading suspicious “cleaner” apps.

📍 OpenAI CEO Sam Altman said that AI technology could usher in a global “fraud crisis.” He said that “AI has fully defeated most of the ways that people authenticate currently, other than passwords.”

📍 French President Emmanuel Macron and his wife Brigitte sued right-wing podcaster Candace Owens, alleging that she “endorsed, repeated, and published a series of verifiably false and devastating lies about the Macrons,” including that Brigitte was born a man. Read the 219-page complaint here.

📍 A buffet of Meta news: it was fined just over half a million USD by Taiwanese authorities for inadequate ad transparency; the company announced it will not sign on to the EU’s AI code of practice; it settled a privacy suit related to the Cambridge Analytica scandal just before Mark Zuckerberg and other key figures were due to testify; and Instagram announced that search engines will soon begin indexing and displaying public photos and video from the platform.

📍 Elon Musk said that xAI plans to make Baby Grok, “an app dedicated to kid-friendly content.

📍 404 Media reported that a UK intelligence firm run by Aidan Raney, a 23 year-old whose profile was growing in the OSINT community, has been selling access to personal data that was exfiltrated by infostealing malware.

📍 Google introduced Backstory, “an experimental artificial intelligence (AI) tool that surfaces information and helps people learn more about the context of images seen online.” The tool, which has not been publicly released, analyzes whether an image was generated with AI and offers information about where it may have appeared online. You can sign up to be a tester at the bottom of this page.

📍 Henk van Ess launched a test version of detectai.live, a free tool that helps detect whether an image is AI-generated. I like that the tool explains its conclusion, encourages the user to run reverse image searches, and provides an option to review additional detail about how it performed its analysis.

📍 OSINT Me published, “OSINT tools and techniques for research on Reddit.”

📍 OSINT Intelligence Terminal is a name/people search tool. (Via Logan Woodward)

📍 Reversely.ai is a reverse image search tool that says it “goes beyond merely finding matching images” and “unlocks hidden connections.”

📍 Last week’s Briefing shared a guide to AI red teaming from Microsoft. A reader replied to suggest a similar resource from UNESCO, “Red Teaming artificial intelligence for social good.” Thank you, Alice!

📍 The Pulitzer Center is hosting a free workshop on August 26, “An Editor’s Guide to AI.” Apply at the link, as attendance is limited. The Center is also holding a free “Introduction to Reporting on AI” workshop on July 30, which will be delivered in Mandarin.

📍 Indicator’s next members-only workshop is August 1 at 12 pm ET. Craig will dive into some of the tools and techniques outlined in our recently published Indicator guide to connecting websites together using OSINT tools and methods. Please upgrade now to attend. The workshop will be recorded and members get access to the video, slides, and a transcript. (Existing members will receive an email with the workshop link soon.)

📍 FactLink, a digital investigations and media literacy organization in Taiwan, published a report (in Chinese) about YouTube channels with AI-generated content that “create fake video stories that are picked up by Taiwan’s mainstream media, amplifying AI-driven misinformation.” Read an English summary here.

📍 CheckFirst published, “Decoding Secrets Through Symbols: How Military Insignia Revealed Russia’s Hidden SIGINT Network.” They studied Russian military decorations to reveal new information about one of the FSB's most secretive units. Also read this great Q&A with the organization’s CTO, Amaury Lesplingart, about CheckFirst’s approach to building tools.

📍 The Institute for Strategic Dialogue published, “Online radicalization and the nexus to violence in the US: 2024 year in review.”

📍 Researchers at Wharton tested how ChatGPT responds to requests that use “classic persuasion principles like authority, commitment, and unity.” They found that the approach “can dramatically increase an AI’s likelihood to comply with requests they are designed to refuse.”

📍 The World Economic Forum published, “Rethinking Media Literacy: A New Ecosystem Model for Information Integrity.”

📍 Uh oh: “Users who encountered an AI summary clicked on a traditional search result link in 8% of all visits. Those who did not encounter an AI summary clicked on a search result nearly twice as often (15% of visits), ” according to a new Pew Research Center study that looked at Google’s use of AI Overviews in search results.

Want more studies on digital deception? Paid subscribers get access to our Academic Library with 52 categorized and summarized studies:

Indicator is a reader-funded publication.

Please upgrade to access to all of our content, including our how-to guides and Academic Library, and our live monthly workshops.