Dispatch: Debunking the hoaxes surrounding the Adamuz train disaster
Essential verification lessons from the Córdoba disaster, plus WebRecon tutorials and UNESCO’s updated investigative manual
Your weekly briefing on digital investigation, OSINT, and the information war.
⚡ The 30-second brief
The debunk: Debunking the hoaxes surrounding the Adamuz train disaster
New tech: WebRecon for comprehensive crawling, data extraction, and analysis
OSINT stack: Reconnaissance tools for data extraction and vibecoding
🔦 In the Spotlight
Fact-check: AI-generated images of the wreckage were circulating after a deadly train crash in southern Spain. Read the full breakdown from Spain’s VerificaRTVE and 3CatInfo.
Monitoring: Donald Trump’s continued threats to seize control of Greenland sent European leaders into a tailspin. Denmark and other allied countries sent troops to Greenland to participate in military exercises. With the situation escalating politically, we are monitoring for opportunities to add intelligence through open-source investigations.
🖋️ The Lead: Debunking the hoaxes surrounding the Adamuz train disaster
A crash between two high-speed trains on January 18 in Adamuz, Córdoba, left at least 41 people dead and injured more than 100 people. Once emergency services began to process the scene, official images began to emerge showing the extreme wreckage that was left along the track in the wake of the collision.
Getting a clear assessment of the situation on the ground was complicated by the spread of AI-generated images among the genuine footage. And it wasn’t just images: false claims were also circulating before investigators had even had time to make a full assessment of the circumstances surrounding the crash.
Debunking the fakes:
Two widely circulated images were found to be AI-generated by comparison against existing official footage from the scene, showing that the trains were not positioned exactly as they were in reality. (One of these AI pictures was even used by a newspaper in its coverage of the crash.)
A fake message circulated asking the public to help at the scene of the crash, and to come wearing high-visibility vests. The Regional Government of Andalusia confirmed that this was a hoax, and asked civilians not to come to the site.
Posts claimed that Spain gave €247 million to improve Morocco’s railway network, but that was misleading. In reality, Spain did give Morocco (and Uzbekistan) funding, but in the form of repayable loans. In Morocco’s case, the loan was subject to a Spanish company being awarded a contract for the job.
Social media posts claimed the trains were speeding, but the national railway company RENFE said the trains were travelling at 205 km/h and 210 km/h, on a section of track with a 250 km/h speed limit.
— Maria
⚙️ The intelligence stack
🔍 WebRecon
WebRecon is an advanced web reconnaissance tool designed for cybersecurity professionals, penetration testers, and security researchers. It automates the process of gathering intelligence from target websites through comprehensive crawling, data extraction, and analysis.
💡 Quick tip: When conducting initial reconnaissance on a high-interest target, avoid triggering security alerts (like WAF blocks) by running a shallow, filtered scan through a proxy.
The Command:
python3 WebRecon.py [TARGET_URL] --max-depth 1 --no-images --proxy socks5://127.0.0.1:9050📍 Data discovery via vibecoding: Using AI to find hidden data sources (via Digital Digging with Henk van Ess)
📍Mastering the SIFT method: A practical manual on using lateral reading and search operators to verify digital leads and protect the chain of custody. (via Eurovision News Spotlight)
📍Suspect and car investigations: A technical walkthrough detailing the methodology for linking isolated data points—specifically suspect identities and vehicle movements. (via UK OSINT Community)
📍Kling AI: Generating a synthetic person using Kling’s new motion control engine (via Emmanuelle Saliba)
📍Domain search: Domain data for research purposes without any fee or paywall (via ip.thc.org)
📡 Signal analysis
Why Grok restrictions won’t stop society’s latest AI scourge
Kate McDonald explores how X’s new geoblocking and image-editing restrictions for Grok fail to address the root issue, as “Spicy Mode” and standalone app access continue to facilitate the generation of non-consensual deepfakes. (via RTE Clarity)
Before we go…
UNESCO has published the second edition of Story-Based Inquiry: A Manual for Investigative Journalists, offering a comprehensive update of its 2009 global reference handbook.
This fully updated second edition brings new research, innovative practices and insights from global investigations. (via UNESCO)