Fifty Shades of OSINT: from Investigative journalism to economic intelligence

OSINT and AI in Journalism: The Sword and the Shield

As in many other fields, artificial intelligence is transforming journalism—and the information ecosystem more broadly—along two axes: offensive and defensive. It amplifies investigative capabilities while simultaneously making manipulation techniques more sophisticated.

Damien Brunon, Head of Education at ESJ Lille, illustrates the positive effects of this transformation on journalism training. His school has integrated what he calls “computer-assisted journalism,” after seeing how effective these new methods can be. He cites a striking example: the automated analysis of a file containing around 15,000 candidates in local elections revealed that “among the 119 candidates over 90 years old, 80% were Front National.” This triggered a field investigation which ultimately led to a court conviction for “abuse of weakness.” Some candidates hadn’t even realized their names were on the lists.

To highlight how AI changes the equation, Brunon notes that “in just one afternoon, with a small GPT, you can carry out tasks that previously took weeks of work,” reminding us that “the most precious resource is journalists’ time.”

But this same accessibility also benefits disinformation producers. Eliot Higgins, founder of Bellingcat, dissected these mechanisms during his talk. He describes a structured ecosystem of “official state actors,” “private interest actors,” and “true believers.” According to Higgins, “these people use manipulated images to trigger algorithms and generate engagement,” creating economic ecosystems of manipulation.

This evolution has given rise to new journalistic specializations. Shayan Sardarizadeh, member of BBC Verify, presented methods tailored to verifying AI-generated content. His team, created in 2023, brings together nearly 60 OSINT investigators, making it “the largest open-source investigation unit within a newsroom worldwide.” On stage, he submitted several images and asked the audience to vote: real or AI-generated? The results were telling—a majority guessed wrong multiple times. “Even for specialists,” admits Sardarizadeh, “the boundary between real and artificial is becoming extraordinarily hard to draw.”

OSINT: Mapping a Mafia

Two students from the École de Guerre Économique (EGE), Thomas Chauveau and Paul Kremer, presented in Lille an investigation into the influence of the Nigerian mafia known as “Black Axe” in Edo State. Their goal was not to retell the group’s history, but to demonstrate “how, starting from open-source investigative reports, social media, videos, and press articles, we managed to establish links between different entities.”

Their research connected local political figures to known Black Axe members. For instance, Tony Kabaka, an active supporter of the 2024 election’s winning party, publicly declared his affiliation with the movement. On social media, he often appeared alongside the elected governor. “That gave us a first link between the governor and a Black Axe member,” one student explained.

Beyond politics, their work revealed connections with the cultural world. By tracing the networks of certain musicians, they identified troubling overlaps with individuals involved in money laundering. “We wondered if the Pan-Africanism some singers champion wasn’t just a cover for Black Axe activities,” they said, pointing to the instrumentalization of activist rhetoric for criminal purposes.

To carry out this investigation, the students used a variety of methods: analyzing online videos, reviewing local newspapers, cross-checking with expert reports, and closely observing Facebook, TikTok, and Instagram activity of political and cultural figures. “We were able to gather a maximum of information, allowing us to map out an entire interconnected network across multiple spheres,” they concluded.

OSINT: revealing corporate vulnerabilities

The testimony of French pharmaceutical group Guerbet showed how OSINT also plays a strategic role in the private sector. Competing against rivals ten times larger and in heavily regulated markets, the company had to professionalize its monitoring. “For us, OSINT is first and foremost about meeting intelligence and information needs,” explained its compliance manager.

Examples shared illustrated the diversity of risks. In Russia, OSINT revealed that a proposed German distributor during the war and sanctions was actually a spin-off of a Russian supplier already tied to competitors. “On the company website everything was in Russian, which wasn’t very reassuring,” recalled the speaker, who ultimately identified the same actor in a trade fair photo, selling both their products and those of rivals.

In Iran, open-source research—including combing through local press in Farsi—uncovered a money-laundering scheme worth several million dollars: “Without knowing it, we had been approached by a sprawling organization already accused of corruption,” Guerbet’s team said.

Beyond fraud detection, OSINT also serves as a commercial compass. By tracking patents, public drug registries, or import-export data, the teams learned to estimate competitors’ production capacity. The process may appear artisanal—using business registries, sanctions databases, local press, social media, satellite imagery, and even promotional videos of factories—but it becomes crucial when guiding a board of directors or a subsidiary’s strategy. “What really matters is putting facts on the table so everyone is aligned,” stressed the speaker, emphasizing how verified information enables clear decision-making.

As Nicolas Katz summed it up: “We use OSINT with modest means, but it protects the company and provides negotiation keys that are essential for sales teams.”