Briefing: AI chatbots gone wild and tips for investigating X and VSCO

Our weekly Briefing is free, but you should upgrade to access all of our reporting, resources, and a monthly workshop. If you’re already a member, thank you for your support!

Craig spoke to the CEO of GeoSpy about the present and future of using AI for geolocation. The Q&A was quoted in a 404 Media article about the LAPD’s interest in using the tool.

Alexios wrote about 94 TikTok accounts with 1.7 million followers that were using AI avatars of real journalists to spread Spanish-language misinformation. The story led him to a Brazilian WhatsApp group that serves as a marketplace for buying and selling monetized social media accounts. TikTok banned all of the accounts we wrote about.

📍 Jeff Horwitz of Reuters has two alarming stories about Meta’s AI chatbots. He reported on internal documents that said the tools could have “sensual” conversations with kids and provide false information, so long as it included a disclaimer. He also tells the story of a cognitively impaired New Jersey man’s tragic infatuation with a “flirty” Meta AI replica of Kendall Jenner.

📍 Three Ghanaian men were extradited to the US after the Department of Justice accused them of conducting romance and business email compromise scams that netted them as much as $100 million over seven years.

📍 The chairman of the Hong Kong Bar Association is advocating for a “targeted law” against deepfake porn.

📍 A group of 22 pro-Beijing Facebook accounts spread a deepfaked video that allegedly showed three exiled Hong Kong activists discussing their fear of getting extradited to China.

📍 TikTok is looking to layoff roughly 40% of its trust and safety team in Germany so it can replace them with AI and outsourced labor, according to The Guardian.

📍 The Ottawa Citizen revealed that the operator of a network of MAGA bots on X is a 24-year-old resident of the Canadian capital. Saihajpreet Singh said the accounts, which reportedly cost him thousands of dollars a month to run, showcase the AI-infused "Real-Time Narrative Control” platform he’s eager to sell to customers.

📍 AI newsbites: The Register found that some UK real estate agents are using AI to make listings more appealing. Truth Social’s new AI Search appears to have gone woke by stating that the 2020 election wasn’t stolen, among other MAGA blasphemies. A man who turned to ChatGPT for advice on reducing salt intake gave himself a rare psychiatric disorder.

📍 Meta is giving new tools to brands to help them report scam ads to the company.

📍 Impostors are impersonating TechCrunch reporters to extract proprietary information from targets.

📍 Data journalism instructor Paul Bradshaw published a 19-minute YouTube tutorial, “The 3 chords of data journalism: sorting, filtering and percentages.”

📍 TraceTracks is a Python tool that “discovers YouTube videos based on SoundCloud tracks.” (via The OSINT Newsletter)

📍 My OSINT Training added a new free bookmarklet that can extract the date of when an X profile banner image was uploaded. Read our detailed guide on how to use MOT’s bookmarklets to help with your investigations.

📍 Colin Crowden outlined the OSINT techniques you can apply to photo-sharing platform VSCO. He writes that “unlike Instagram, it doesn’t strip photo metadata consistently, meaning each uploaded image can retain crucial EXIF data.”

📍 Ben Heubl shared tips for accessing cached webpages on Yandex and Bing. (Google cache RIP.)

📍 You can now report an an entire Google profile if you believe that it’s submitting “inappropriate” content/reviews in Maps. The company previously only allowed users to report specific reviews for being fake or otherwise violative. More info here. (via Joy Hawkins)

📍 The UK House of Commons Science, Innovation and Technology Committee published a new report on “Social media, misinformation and harmful algorithms.” Recommendations include visible labeling of all AI-generated content and algorithmic demotion of fact-checked misinformation.

📍 Rutgers professor Kiran Garimella reflects on his research agenda now that the climate for those studying misinformation is “unwelcoming.” He acknowledges that “this pivot is a retreat” but hopes to continue asking “meaningful questions about society, technology, and human behavior.”

📍 Internet law scholar Daphne Keller is not a fan of the proposed NO FAKES Act, a bill intended to expand intellectual property rights to cover AI replicas of real people. Keller writes on Bluesky that the law is too broad and risks leading to abusive takedown requests.

📍 Information warfare expert Renee DiResta writes that the “Gabbard Files,” which claim to show that Obama administration officials manufactured intelligence to discredit Donald Trump’s 2016 victory, are part of a well-trodden formula: “take a pile of documents, strip them of context, pace the release for maximum drama, and narrate it with insinuations that travel faster than any correction.”

Want more studies on digital deception? Paid members get access to our Academic Library with 55 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 to our live monthly workshops.