Why "Is This AI?" Is the Wrong Question
And What Analysts Should Ask Instead.
Hi and welcome, the Editorial Team here.
AI detection tools dominate the conversation about verifying synthetic content. This week we argue that detection is the wrong starting point, and that the questions analysts should ask are older, harder, and more productive.
We use the Akhbar Enfejari case from the 2026 Iran conflict to show what provenance-first analysis actually looks like in practice.
Tips and news as usual below.
Missed last week?
Before we get into it, here is last week's tradecraft and industry news roundup:
OSINT Tradecraft Tips & Industry News
Everything you need to know from the past seven days.
An excellent list of tools that allow users to access social media content without creating an account or logging in, curated by Aleksandra B. | see here
Damien Leloup explains how journalists can investigate online scams in a #GIJNAcademy masterclass video | watch here
Micah Hoffman reported that Instant Data Scraper quietly changed ownership to Flavr Technology LP with little transparency, prompting concerns and a recommendation to uninstall the extension | check out here
Book recommendation: We came across The SAR Analyst's Handbook by Kyle McCloud. We have not read it, and this is not a paid promotion, but it looks like a practitioner-written, practitioner-focused guide to SAR analysis worth knowing about | explore here
Grok created a fake French citizen complete with ID and notes that video verification is becoming ineffective, according to Pedro J. | view here
Citizen Lab's new research exposes two sophisticated surveillance actors exploiting global telecom infrastructure, marking the first time researchers have directly linked combined 3G and 4G network attacks to mobile operator infrastructure | read here
A Pew Research Center report found that 26% of web pages published between 2013 and 2023 are no longer accessible, but a new Internet Archive study shows that the Wayback Machine has restored 16% of lost pages and preserved 56% before they disappeared | read the report
The explosion of OSINT, UAS, and AI-driven analytics has flooded the US intelligence enterprise with more data than analysts can process, and fixing it requires hybrid human-machine workflows rather than purely technical solutions, says Jared Martin in Smallwarsjournal | read here
Orchestrated autonomy will replace AI assistants in intelligence work — networks of agents will monitor, analyse, and surface insights continuously, removing the need for analysts to initiate queries, writes Rep. Scott Perry in Federal News Network | read here
Why “Is This AI?” Is the Wrong Question — And What Analysts Should Ask Instead
