In partnership with 


The open source Justice and Accountability Unit is a partnership with Bellingcat. At GLAN, it is led by Dearbhla Minogue and Siobhan Allen (executive team), with input from Professor Yvonne McDermott and Dr. Ioannis Kalpouzos (legal action committee). The lead investigator at Bellingcat is Nick Waters, who is supported in the design of the programme by Hannah Bagdasar and Charlotte Godart. 

In the area of data protection, the project also benefits from the advice of Monika Sobiecki of Bindmans LLP, and Alex Bailin QC and Ben Silverstone of Matrix Chambers. Input on the criminal procedure aspects of the law of England and Wales relevant to the methodology was provided by Emilie Pottle of Temple Garden Chambers and Daniel Robinson of Red Lion Chambers.  GLAN and Bellingcat also thank Helen Malcolm QC, Andrew Cayley QC, Joshua Kern and Shina Animashaun for their invaluable participation in our mock hearing exercise. The project’s procedures also benefit from an independent external review panel of experienced international criminal law practitioners.GLAN and Bellingcat are grateful to Mnemonic and HURIDOCS, whose preservation and databasing systems form a key component of the project.


About the PRoject

The fair and successful prosecution of atrocity crimes requires reliable evidence establishing not only that an event took place but that a specific perpetrator was individually criminally responsible. Efforts to prosecute such atrocity crimes can be hampered by the absence of such evidence, for example where official investigators cannot reach a given location promptly, where witnesses’ memories fade and change, or when there is simply not enough first-hand evidence. Even as universal jurisdiction prosecutions proceed and the international criminal court widens its caseload, perpetrators’ reliance on impunity seems to prevail, with fresh atrocities taking place regularly. 


Meanwhile, the advent of social media and smartphones has combined to result in an unprecedented volume of online videos, and inevitably some of those videos can evidence key aspects of alleged atrocity crimes. This can include footage filmed by perpetrators for propaganda purposes, or footage captured by horrified passers-by or survivors who want to document and expose the harm they have witnessed. A video of a commander executing captured men could amount to evidence of murder; footage of an airstrike impacting while desperate civilians attempt to rescue a young casualty could amount to evidence of an attack in contravention of the laws of armed conflict. 


However, courts are unused to dealing with digital evidence whose origins are unknown, and there is a significant risk that the full potential for this content to be harnessed for justice could be missed or delayed. GLAN and Bellingcat have been breaking new ground in the fight against impunity for atrocity crimes through ensuring that open source investigations can give rise to admissible, reliable evidence. Through years of collaboration, we have developed and tested a step-by-step methodology for investigators to follow which will ensure that courts can put aside any skepticism and place appropriate trust in open source information. The genesis for the project was an expert workshop convened in 2018 at which we began considering the potential for open-source information to be used as evidence in legal proceedings. The methodology, which is akin to a standard operating procedure, takes into account best practice guidelines set out in the Berkeley Protocol on Digital Open Source Information.

The collaboration has given rise to a separate workstream within Bellingcat for accountability-focused investigations - this is known as the Justice & Accountability Unit. 


our current projects

YEMEN Our open source project began as part of our organisations’ commitment to pushing for accountability for aerial attacks causing grave civilian harm in Yemen by the Saudi/UAE-led coalition.


Drawing on Bellingcat’s investigations into airstrikes in Yemen, GLAN has analysed the contribution made by online open source information where an assessment of compliance with international humanitarian law (IHL) is required. This study, conducted in conjunction with Dr Ioannis Kalpouzos of GLAN's Legal Action Committee, is soon to be released. For more on our work in relation to open source information and Yemen, see our Yemen case page


UKRAINE: In response to the invasion of Ukraine by Russia, GLAN and Bellingcat is preparing to launch a set of investigations into alleged atrocity crimes taking place there. These investigations will follow the methodology and their findings will be made available to any national or international prosecutors who are gathering evidence of alleged crimes. The evidence will be arranged in an analysis database which can be searched and filtered. Preservation will be done by Mnemonic, an independent third-party organisation maintaining an archive of digital content from Ukraine, as it has done for Syria, Yemen and Sudan. On 10 and 11 March 2022, GLAN and Bellingcat gathered at Bindmans LLP in London to begin planning for this long-term project, and discussed all aspects of the investigation, preservation and databasing process, including data protection and privacy concerns, so that the evidence could have the greatest possible chances of successful introduction in criminal courts as evidence. Click here to find out more.



Watching an incident actually occurring is a vastly different experience to hearing a description of it afterwards from someone else who was there. This is why videos and photographs have played a key role in criminal and civil trials since the technology became available, in contexts as varied as insurance disputes to atrocity crimes. In the vast majority of cases, the authenticity of a video - that is, the question of whether it is genuine - is proven by introducing evidence of provenance. Provenance refers to the origins of an item - its creation. 

For example, journalist Ed Vulliamy appeared as a witness before the International Criminal Tribunal for the former Yugoslavia and his infamous video of the prisoners at Omarska camp was admitted as evidence during his testimony as he was able to speak directly to what it depicted. But how does the court react to a video which has appeared online, with no evidence of its provenance? Since 2018, GLAN and our partners at Bellingcat have been merging legal and investigative experience to answer this question so that videos uncovered and authenticated by Bellingcat could be admissible as evidence in a criminal trial. A court will need to satisfy itself of a range of factors in order to conclude that it is fair to admit an online video into evidence against a defendant in criminal proceedings.



This is the big question. A video is considered by a court to be a document, and as such, it must be authenticated like all other documents submitted as evidence. But what does it mean for an item to be authentic?  In general, the concept translates into whether the item has been forged or faked. It is much easier to prove authenticity when a reliable witness, such as a crime scene photographer, can come to court and testify that he or she is the creator of the image and that it has not been altered since its creation. Similarly, a CCTV or satellite image can be authenticated by a person with knowledge of the technical system that captured the images. 


However, when a video has been found online, no such evidence from its creator is available. This is where the skills of an online open source video analyst can be critical. Open source investigators can confirm authenticity by performing verification analysis on videos through techniques such as geolocation, chronolocation, cross referencing, and checking for internal consistency, amongst others. They can ultimately then take a view on whether the item, in particular in combination with other items such as satellite imagery or other online videos showing the same event or location, is genuine. 


While this kind of verification is common for journalistic purposes, the key legal question is whether such a piece of video is reliable enough for a court . In order for a court to be in a position to accept the item’s authenticity, an online analyst would have to give their opinion in the form of so-called ‘expert evidence.’ However, given that this is a relatively new field without formal academic qualifications, would a court consider an open source analyst an expert? And, crucially, can a criminal court deem an item authentic without any evidence of provenance? We explored these and other questions through a mock hearing exercise



In many circumstances, a private organisation would not be the only source of a piece of open source content, because an official investigation - even if handed the evidence as a referral or criminal complaint - would carry out its own online investigation and uncover the content itself. However, there could be circumstances in which a private investigating organisation (such as Bellingcat) is the only source of an item, for example if the item was taken offline before official investigators found and preserved it themselves, or in the case of a private civil or criminal prosecution. In that case, the process by which the evidence was obtained - i.e. the open source investigation - would come under scrutiny. This is because the court will need to assess whether it is fair to introduce the evidence against the accused, taking into account the manner in which it was obtained. If no information exists as to this process, it increases the likelihood that the defendant will be unfairly prejudiced by the introduction of a piece of evidence, leading to the exclusion of, or attribution of low weight to, the item. This is where the GLAN/Bellingcat methodology comes in. 



When video is being introduced as evidence of a crime, it is crucial that the potential for omission is taken into account. An obvious example is a video being taken of the aftermath of an airstrike which has damaged a civilian car, but which omits to capture a military tank that was parked next to the car. 

This is partially an issue for investigators and partially an issue for lawyers wishing to introduce the evidence. Investigators should always endeavour to find the earliest and most complete version of an online video, so that they can check whether they are being shown the full context. At the end of the day, however, investigators and analysts can only comment on what they can see, and cannot speculate as to what the videographer may have missed out. Lawyers seeking to introduce video as evidence where omission may be an issue should be conservative as to the claims being made about what elements of the crime the video is said to prove. 



The theoretical potential for the staging of items in an otherwise authentic video complicates things further. For example, a bomb remnant linking a real attack to a particular party could be placed at the scene in order to garner outrage, or civilian items such as toys could be added and photographed to enhance the propaganda value of an attack. However, this possibility should not deter courts from placing any value on content depicting moveable items. Any moveable items should be treated with a level of caution appropriate to the setting and context.



A court will need to be assured that the version of the video it has access to is identical to the version taken from the internet by the investigator in question. Preservation and maintaining chain of custody are very important issues around which a range of expert organisations are doing essential work. See, for example, the work of Mnemonic, the independent organisation which performs preservation of content Bellingcat uncovers in Yemen and Ukraine. 

Of course, in an ideal world, this chain of custody would exist going back further than the item’s appearance online to the moment of its creation, but by the very definition of open source information, that is not possible. In addition to the creator being unavailable to attend court, social media platforms remove metadata which could provide information as to the circumstances of the item’s creation. It is therefore necessary to authenticate the video using alternative methods of analysis, as described above, and thereafter to ensure chain of custody is maintained once a copy has been taken offline.  


Hackathon participants collating and verifying evidence across a number of controversial airstrikes while following a methodology designed by GLAN and Bellingcat


Online open source investigation can result in the discovery of unmanageable volumes of media, which can take many formats. The inability to track high volumes of evidence and incidents and view multiple items simultaneously can hamper efforts to move swiftly and cohesively as a team. This can also make collaboration with external partners difficult and cumbersome. To address this, we have developed a dedicated evidence analysis platform in conjunction with HURIDOCS (Uwazi). 


Once the investigations team uncover information they consider could amount to relevant evidence, it is preserved, either on an ad-hoc basis using the Chrome application Digital Evidence Vault, or by the independent organisation Mnemonic. After that, the information is entered into a custom-built analysis database so that the evidence can be collated, filtered and understood according to parameters relevant to criminal litigation. This database can then be accessed by national or international prosecutors or used as a basis for private litigation. 



analysis database

in partnership with