The Open Source Intelligence (OSINT) profession is facing arguably its greatest challenges to date. Online information volumes are expanding faster than ever before and data is increasingly fragmented, existing across a growing number of platforms and pages. To give a sense of scale, there are around 250,000 new websites registered every week, and the average social media user operates across 6.5 platforms. This rapidly evolving environment allows misinformation and disinformation to spread at speed and scale.
AI is of course fuelling these trends through mass content generation – 2.5bn messages are sent to ChatGPT per day as of July 2025 – and deepfake technology, enabling bad actors to enhance their deception techniques. All of this makes it increasingly challenging for investigators to detect and verify genuine information.
At the same time, OSINT investigators are expected to complete investigations faster and derive deeper insights. Yet the reality is manual OSINT workflows are now unsuited to the scale of online data, especially where speed is required. No longer can an analyst spend extended periods of time performing deep analytical work on a single event before moving on to the next one – after 24 hours, that event will already be old news.
Because of this, AI has become a vital tool for redefining traditional OSINT workflows and empowering investigators to manage the vast amount of publicly available data needing to be analysed with faster turnaround times and higher accuracy. But understanding where AI can elevate OSINT tasks and where human judgement remains critical is essential to its effective and safe use. Without this understanding, its benefits turn into risks.
Also Read: CIO Influence Interview with John Elliott, Cybersecurity Author Fellow at Pluralsight
The strengths and weaknesses of AI as a tool for investigations
If companies are to understand how AI can help overcome their OSINT challenges, then they need to know its strengths and weaknesses. One of AI’s main investigation capabilities is automating publicly available data collection and triage. When performed manually, this process can consume a large chunk of an investigator’s time on a case – up to 70% – with workflows often reliant on searching through one source at a time. AI can automatically gather and sort relevant information from all of these sources simultaneously, reducing the cognitive load on the analyst.
One of the key tasks for an investigator is entity resolution, where they have to link various online data, profiles and records to the same real-world identity. Another one of AI’s strengths is extracting entities and patterns to surface these connections to an analyst. Finally, AI can succinctly summarise and process large datasets, helping investigators leverage the abundance of information available online. With this information consolidated and presented in a digestible way, analysts have far more time for contextual interpretation and spotting patterns in the data, using AI insights to inform their analysis.
What AI cannot do is conduct end-to-end investigations without human input. Accountability for any decision an AI system makes still rests with a human. If someone relied on AI to conduct an entire investigation and something went wrong, the resulting harm, and the responsibility for it, would fall to the human, not the machine.
Just as crucially, AI cannot replace human insight. It excels at automating repetitive tasks, but it struggles to interpret context and exercise judgement, meaning it cannot, on its own, produce intelligence. It is also prone to hallucination and to presenting incorrect information with confidence.
Redefining tasks: how AI can change OSINT workflows
So, what could an investigation with humans and AI working side-by-side look like?
The first step is setting the direction of the case. An investigator defines the goal through a specific prompt. For example, “Acting as an OSINT Investigator, conduct a due diligence investigation into this Company. Identify adverse media, risk factors, and any other factor that may influence an M&A decision”. AI then accelerates the scoping and exploration of searching public data sources in line with this goal. While the technology enables rapid, automated data collection, humans judge the proportionality of this. In other words, they determine the sources that are appropriate, relevant and ethical to access. AI is then able to reduce the time humans need to spend on manually filtering, mapping and verifying this data.
The next step is analysis. While human analysis is critical, AI can enhance insights by surfacing risk factors and patterns in the network. This is particularly crucial for large, complex networks, as manual analysis takes hours and could mean something important is missed.
Once analysis has taken place, creating and disseminating a report (like a SAR) is crucial for ensuring information reaches the relevant enforcement agency in a clear, accurate and timely way. AI can assist investigators in building these reports efficiently, while the human is responsible for drawing conclusions and creating intelligence.
What’s vital throughout this workflow, however, is that there is always a human-in-the-loop or on-the-loop, as appropriate. What do I mean by this? The term is used widely, but for OSINT investigations, it means that, if necessary, a human can intervene in any step performed by AI. This depends on investigators having visibility into where data has come from, how the system has produced an output and why it has done so.
Maintaining control: the central role of the human analyst
As we’ve seen, AI can do the heavy lifting of repetitive data collection and processing. But unlike conversations in other fields around AI displacing roles, it’s not going to take over the role of an analyst, and nor should it. An AI model doesn’t have the several years’ experience an OSINT expert might have built up in a specific field or area. It can collect more information than they can, but it won’t have the contextual knowledge they do or the insight to generate reliable intelligence. This is why the role of a human analyst is augmented by AI, as it gives more space for experience and talent to prosper.
Humans have a fundamental role to play in maintaining accountability in decisions, in protecting data and privacy, in creating transparency around the methodology used, and in mitigating bias and hallucinations, problems that will continue to persist with AI. Rather than seeing AI as a revolution, we should view it as an evolution that is simply accelerating existing processes, a progression that is needed to match the speed, volume and variety of information in the online space.
About The Author Of This Article
Chris P. is Head of Intelligence at Blackdot Solutions
About Blackdot Solutions
Blackdot Solutions are the creators of Videris, the most powerful OSINT investigations software for fighting serious, organised and economic crime.
Catch more CIO Insights: How Are CIOs Aligning Technology with Workforce Agility?
[To share your insights with us, please write to psen@itechseries.com ]
