Open-source intelligence, commonly known as OSINT, is developing from a specialist investigative discipline into an increasingly important capability for journalists, law-enforcement agencies, defence organisations and intelligence professionals.

Recent developments reveal several forces shaping the sector: artificial intelligence, technical infrastructure analysis, government procurement, satellite imagery, data governance and professional training.

AI is transforming large-scale OSINT analysis.

One of the most significant developments is the use of artificial intelligence to help investigators process volumes of online material that would be impractical to examine manually.

In June 2026, the Reuters Institute published an account of how BBC Eye developed a multi-agent AI system called Haystack to support an investigation into Russian nationalist groups.

The system gathered approximately 10,000 social-media posts from more than ten groups and produced around 55,000 content assessments. It was used by a team combining OSINT specialists, reporters and computational journalists.

Read the full article: How BBC Eye built a multi-agent AI system to sift through ten thousand Russian social media posts.

This case illustrates one of the most credible uses of AI in OSINT: accelerating collection, classification and prioritisation without removing human investigators from the process.

AI systems can help identify recurring names, locations, phrases, accounts and relationships. They can also group similar material and direct investigators towards content requiring closer examination.

However, an automated assessment is not the same as a verified conclusion. Investigators must continue to account for false positives, incomplete context, translation errors and model bias. The usefulness of AI therefore depends on the quality of the investigative methodology and human oversight surrounding it.

Technical infrastructure can expose hidden relationships.

OSINT investigations are not limited to social-media posts, photographs and videos. Digital infrastructure can also provide evidence of relationships that organisations do not publicly acknowledge.

In June 2026, Bellingcat reported finding several technical links between the video news agency Viory and Ruptly, a branch of the sanctioned Russian media organisation RT.

The investigation identified indicators including shared IP addresses, a Viory-linked website using a digital certificate registered to Ruptly and the apparent transmission of website-performance data from Ruptly to Viory. Both organisations denied having a connection.

Read the investigation: Tracing Digital Links Between Viory and Ruptly.

The case demonstrates the value of technical OSINT in attribution and network analysis. Hosting records, certificates, archived websites, tracking codes and other digital artefacts may reveal connections that are not visible through conventional reporting.

Such evidence must still be interpreted cautiously. Shared infrastructure can indicate a relationship, but it does not automatically establish common ownership or control. Strong attribution normally requires several independent forms of corroboration.

OSINT is becoming part of formal public-sector infrastructure

The institutional use of OSINT is also becoming more visible through public procurement.

BlueLight Commercial has established a multi-supplier framework for software supporting open-source intelligence and social-media screening. The procurement notice states that the framework has a four-year term and an anticipated commencement date of 9 April 2026.

View the procurement notice: BLC0288 Open Source Intelligence Social Media Screening Capabilities.

The framework indicates that OSINT software is moving beyond isolated or experimental use and becoming part of formal operational procurement for policing and other eligible public-sector bodies.

The UK Ministry of Defence has also previously engaged technology suppliers about its requirements for open-source intelligence capability. A techUK industry-engagement event held in October 2024 provides an example of the MOD’s continuing engagement with the commercial intelligence and technology sector.

View the event information: Ministry of Defence Open Source Intelligence Capability Industry Engagement Event.

Institutional adoption brings additional responsibilities. Public bodies need defensible policies covering authorisation, collection, retention, access, review and deletion. They must also distinguish between proportionate intelligence activity and excessive monitoring of lawful public behaviour.

Satellite imagery is strengthening independent investigations.

Commercial and publicly available satellite imagery remains one of the most valuable resources available to open-source investigators.

Following the June 2026 earthquakes in Venezuela, Bellingcat obtained satellite imagery covering a 14-mile section of the country’s northern coast. Investigators compared post-earthquake images with earlier reference imagery to identify collapsed buildings and assess damage in places including Playa Grande, La Guaira and Caraballeda.

Read the investigation: Satellite Imagery Shows Scale of Venezuela Earthquake Damage.

In a separate investigation, Bellingcat combined satellite images, social-media footage, visible landmarks and digital terrain-modelling tools to geolocate burial activity near La Esperanza Cemetery.

Changes visible in satellite imagery showed that an area of approximately 1.5 acres had been cleared. Investigators compared the imagery with videos and distinctive terrain features to establish where and when the recorded activity occurred.

Read the follow-up investigation: Between Graves and Uncertainty: The Management of the Dead After Venezuela’s Earthquake.

These investigations demonstrate the value of combining multiple source types. Satellite images can reveal physical changes, but social-media footage, drone images, terrain data and local reporting are often necessary to interpret those changes accurately.

Satellite analysis also requires care. Investigators must consider image resolution, capture dates, cloud cover, shadows, seasonal variation and previous activity at the location. A visible change does not necessarily explain why that change occurred.

Privacy and governance are central to responsible OSINT

The expansion of OSINT and AI-assisted analysis is making legal and ethical governance increasingly important.

The UK Government’s Data Landscape Review Report 2024 to 2025, published in March 2026, examines the data-access needs of the online-safety technology sector. It discusses the use of alternative and external data sources, advances in large language models and the need to maintain privacy while reducing dependence on extensive proprietary datasets.

Read the report: Data Landscape Review Report 2024 to 2025.

The fact that information is publicly accessible does not mean that every form of collection, aggregation or analysis is proportionate.

Combining data from several public sources can reveal personal relationships, routines and vulnerabilities that would not be apparent from an individual post. Organisations should therefore apply principles including necessity, proportionality, data minimisation, purpose limitation and controlled retention.

Investigators must also consider whether publishing their findings could expose victims, witnesses, vulnerable individuals or uninvolved members of the public.

Good governance should not be treated as an obstacle to OSINT. It is an essential element of producing intelligence that is lawful, ethical, trustworthy and defensible.

Training is evolving alongside the technology.y

As tools become more sophisticated, the need for structured OSINT training is increasing.

The European Union Agency for Law Enforcement Training, CEPOL, has included a 2026 webinar on emerging OSINT trends and investigative tools. Its stated objectives include using AI-driven tools and large language models for automated collection and analysis, identifying synthetic media and bot networks, and tracing digital footprints while protecting investigators’ anonymity.

View the training information: Emerging Trends and Tools in OSINT Investigations.

The programme reflects the changing skills expected of modern investigators. Practitioners increasingly need to understand not only search techniques, but also artificial intelligence, synthetic identities, automation, digital security and operational anonymity.

Technical ability alone is insufficient. Effective practitioners must also understand source evaluation, corroboration, evidence preservation, legal authority, privacy, operational security and the communication of uncertainty.

The defining OSINT skill is not simply finding information. It is establishing what that information proves, what it does not prove and how confidently a conclusion can be presented.

The future of OSINT depends on trust.

The evidence from these developments suggests that OSINT will continue to increase in scale and strategic importance.

Artificial intelligence will allow investigators to examine larger datasets. Procurement programmes will embed OSINT software within public institutions. Satellite imagery and technical infrastructure analysis will support increasingly detailed investigations. Training programmes will adapt to synthetic media, automated accounts and emerging platforms.

The risks will grow alongside the capabilities.

Poorly governed OSINT can result in mistaken attribution, intrusive monitoring, privacy violations and conclusions based on incomplete or misleading evidence. AI can accelerate an investigation, but it can also accelerate errors when its results are accepted without verification.

The future credibility of the discipline will therefore depend on maintaining a balance between capability and accountability.

OSINT is most valuable when its methods are transparent, its evidence is corroborated and its conclusions are open to scrutiny. Technology may expand what investigators can discover, but professional judgement will continue to determine whether those discoveries are accurate, ethical and useful.