Why multilingual Open-source Intelligence (OSINT) matters more than ever

From tracking Russian troop movements in Ukraine to conducting due diligence in M&A deals, Open-source intelligence (OSINT) is the foundation for understanding risk across industries.

OSINT is the practice of collecting and analysing information gathered from open sources to produce actionable intelligence that informs decision-making.

Paul Wright, Neal Ysart and Elisar Nurmagambet, experts at the Coalition of Cyber Investigators, have shared an authoritative analysis on the challenges of online intelligence collection, particularly around evidential integrity, contextual accuracy and investigative tradecraft, which is both timely and essential.

Having served as Director of Intelligence at the City of London Police, I resonate with their findings. Confidentiality, source protection and operational integrity were not abstract principles; they were the difference between safeguarding life and property or placing people at risk. Wright, Ysart and Nurmagambet articulate that OSINT is not only supplementary, but a primary intelligence discipline that requires the same rigour, cultural awareness and protection measures as any conventional techniques.

Introduction

The authors highlight several issues that should concern every professional investigator today:

1. OSINT is expanding faster than many organisations can adapt

The Coalition’s work frequently emphasises that investigators now rely on a vast array of public-facing data sources spread across global digital ecosystems. Their analyses of OSINT practices underscore the importance of structured methodologies, evidential discipline and contextual interpretation, especially as investigations traverse multiple online environments and jurisdictions.

2. OSINT without cultural and linguistic accuracy is dangerous

A critical theme running through their body of work is the importance of context. As they have noted in various publications, misinterpretation, whether linguistic, cultural or behavioural, can result in flawed assessments, mistaken identities, and investigative blind spots.

This is an issue that deserves far more attention.

The risk of ignoring Non-English Open-source intelligence

English remains the dominant language of the OSINT community, but it is not the world’s most widely spoken language.

Investigators cannot afford to overlook or ignore:

• Local slang and dialects

• Online subculture terminology

• Coded or indirect language

• Regional euphemisms or idioms

• Minority language social platforms

These elements often contain mission-critical signals that do not translate cleanly through generic machine translation tools commonly used by investigators.

Secure tools such as CoPilot can be integrated into investigation and e-discovery workflows to translate content and give up to 95% accuracy. But the greatest risk may lie within the last 5%. A single mistranslation, lost nuance, or misinterpreted phrase can compromise an entire line of inquiry.

In extremist communities, criminal networks and fraud ecosystems, coded language is used deliberately to obscure meaning, mislead outsiders, and communicate sensitive information that only insiders will recognise. Wright and Ysart’s broader commentary on the evolution of online deception and fraud dynamics shows how adversaries increasingly use sophisticated digital methods to conceal activities in plain sight.

Missing these signals due to language bias increases the risk of harm, wastes precious investigative resources and damages confidence among stakeholders.

The operational security problem in OSINT collection

One of the most commonly overlooked risks in OSINT work today is the investigator’s digital footprint.

Every search, click, and translation request leaves traces. Some platforms track user behaviour. Others expose metadata. Many online translation engines retain input text, building datasets whose security and privacy controls investigators cannot verify.

As Wright and Ysart emphasise in their discussions on evidential integrity and tradecraft, investigators must protect not only their findings but also their own visibility, identity and methods.

In sensitive investigations, particularly those involving hostile actors or serious organised crime groups, a careless online footprint risks:

• Alerting subjects that they are being monitored

• Compromising undercover operations

• Exposing human or digital sources

• Leaking details of investigative strategy

This is a threat intelligence leaders understand all too well. As a police Director of Intelligence, my number one priority was protecting my officers and the sources who supplied intelligence products. The rigorous application of field craft is essential, especially in fast, high-risk settings. Advances in Artificial Intelligence (AI) tools for investigators have added powerful new tools to their arsenal, which require careful handling to prevent them from being turned against them.

Secure, multilingual OSINT workflows

This is precisely why modern investigators need tools designed with operational security, linguistic accuracy, evidential reliability and interoperability at their core.

Where GAI Translate Helps

While this is not a sales pitch, I would be remiss not to highlight why Guildhawk works closely with regulated professionals like Tenet Law to develop solutions such as GAI Translate and GAI Small Language Models (GAI SLMs) that address the very challenges Wright, Ysart and Nurmagambet discuss.

1. Secure, anonymous translation workflows
Unlike free online tools, GAI Translate and GAI SLMs ensure text and documents never leave a protected environment, preventing investigative material from being exposed to third parties.

2. Automated anonymisation
Investigator identity, IP data and other metadata can be protected throughout the research process.

3. Human-verified accuracy
AI translation is powerful, but sensitive OSINT often needs verification, and this can now be done within GAI with the first 1-click solution enabling users to access a vetted, security-cleared linguist to:

• Verify accuracy

• Interpret nuance

• Identifies cultural or coded signals, and

• Certify translations for evidential use in civil or criminal proceedings

4. Secure cultural intelligence
When research uncovers odd phrasing, slang, or ambiguous language, expert, vetted linguists provide context that an algorithm alone will miss, exactly the type of cultural awareness required to reveal OSINT blind-spots.

In the world Wright, Ysart and Nurmagambet describe, one where OSINT is increasingly sophisticated, cross-border, and mission-critical, these capabilities are not “nice to have.” They are essential. This was also reflected in the judges’ praise when honouring GAI with the Safety & Security Entrepreneurs Awards, Market Disruptor of the Year award.

Final thoughts

The work of Paul Wright, Neal Ysart and Elisar Nurmagambet is a reminder that OSINT is undergoing a transformation. As the digital landscape becomes more multilingual, more deceptive and more operationally complex, our investigative tools and practices must evolve with it.

For investigators, intelligence professionals and analysts, the message is clear:

• Do add secure, tried and tested translation tools like GAI to your arsenal.

• Do not rely solely on English language OSINT.

• Do not trust generic translation tools with sensitive material.

• Do not underestimate cultural nuance.

• And never compromise operational security in the pursuit of intelligence.

Their insights strengthen the profession, and they underscore why secure, multilingual, culturally aware intelligence workflows are now indispensable in the age of AI.

If we are to protect people, assets, and communities effectively, as I had to every day in my former role as a United Nations Police Investigator, we must ensure our OSINT practices meet the threats we face.

And that starts with understanding the world in all the languages it speaks.

Original article by David M Clarke from GuildHawk

David works closely with Guildhawk's in-house software developers, partners at Sheffield Hallam University, and clients to develop new, secure AI technologies to address risks in current AI translation, including introducing the SSEA award-winning GAI Translate, backed by Innovate UK.

A former police Detective Chief Superintendent, veteran of the Royal Air Force, and former chief investigator with the United Nations International Police Task Force in Bosnia & Herzegovina, he champions Guildhawk’s commitment to helping military veterans find new careers in civilian life.

As the Head of the City of London Police Fraud Squad, he was responsible for designing and leading the National Fraud Intelligence Bureau (NFIB). As Chairman of the independent Fraud Advisory Panel, David established and led the COVID-19 Fraud Watch Task Force to protect businesses during the pandemic. He is a regular commentator on BBC Television and The Times newspaper.