At the closing conference of the ‘Digital Rights: New Technologies for Criminal Law’ project, held in Venice on 12 and 13 October 2026, judges, legal experts and practitioners analysed the implications of the digital revolution for criminal proceedings. Drawing on the discussions held during this meeting, Olivier Chevet and Sami Kodia examine the main challenges posed by digital evidence and the growing use of artificial intelligence in investigations and legal proceedings.
Contribution by Olivier Chevet*
* A judge in the judicial system, currently head of studies and research at the Robert Badinter Institute. He leads work on the ‘Justice and Digital Technology’ programme.
The growing abundance and sophistication of digital evidence are creating new challenges for criminal proceedings.
On the one hand, digitalisation significantly increases investigators’ ability to collect, cross-reference and analyse information. On the other hand, this evidential power reduces the ability of the parties and the judge to understand, verify and challenge the results obtained. This was the central issue addressed at the closing symposium of the ‘Digital Rights: new technologies for criminal law’ project, held in Venice on 12 and 13 October, as well as in the forthcoming report ‘Deliverable D3.4 – Soft Guidelines on Legislative Implementation’. Without claiming to be exhaustive, here are some of the key points that merit attention.
The first of these is the rapid development of digital evidence, extending far beyond cybercrime offences. The second is that, when it becomes dematerialised and digitised, the evidential process is the result of calculations that are sometimes highly sophisticated and opaque, which almost alter its very nature. The third is that this dematerialisation is accompanied by a deterritorialisation that affects the legal structuring of the world around the concept of the sovereign state. The fourth point, posed as a question, concerns the impact on the conduct of criminal proceedings and on the rights of the parties, to the extent that some are questioning whether the traditional form of the trial itself might be disrupted.
The rapid development of digital evidence
The view shared by all those involved in the Digital Rights project is that digital evidence is becoming very common, to the point of being ubiquitous.
The development of digital spaces and the gradual digitisation of entire areas of social and human activity have led to the emergence of purely digital forms of crime, that is to say, offences whose material manifestation takes the form solely of the manipulation of the information space, through the alteration of data storage media. The most obvious examples are offences targeting information systems: data theft and trafficking, denial-of-service attacks, and ransomware. This category may also include online fraud and illicit fund transfers, whether via the largely digitised financial system or through cryptocurrency mechanisms. Offences against individuals are also covered, for example online harassment or the distribution of child sexual abuse material.
This process of digitalisation is also reflected in the increasing use of digital evidence in offences committed in the physical world. This may be because digital devices were used by the perpetrators in the commission of the offence – such as, of course, mobile phones, the viewing of digital content, the use of connected vehicles, digital communication platforms, geolocation tools or the use of digital services. But it also applies when personal digital devices produce traces that may establish involvement, such as CCTV footage or connection logs.
In all these cases, the evidence takes the form of digital traces left in the targeted digital systems, but also in those used to commit these offences. These traces are stored on media such as USB sticks or hard drives. They are also found in the storage spaces of the providers operating these systems. Owing to this characteristic, these digital traces have several important specific features
Is digital evidence of a different nature?
The forms of evidence that might be described as ‘traditional’ are based on various types of recorded statements from individuals with varying statuses – whether they be defendants, witnesses, victims, investigators or forensic pathologists – as well as translations and interpretations. These forms of evidence can be utilised almost immediately by those involved in the trial. By contrast, scientific and forensic forms of evidence rely on the application of knowledge, expertise, and instruments capable of transforming raw traces into evidence. Digital evidence falls into this second category.
The Digital Rights project proposes a very interesting five-stage model[1] of the procedural and technical process involved in transforming raw digital data into admissible evidence:
- the collection of digital data,
- forensic extraction,
- data filtering and processing,
- algorithmic analysis
- and judicial assessment.
The project’s findings highlight the importance of documentation and the integrity of the chain of custody and processing of digital evidence. This is indeed essential, given the increasing complexity of the processes in place, which makes it more difficult for those involved in legal proceedings to understand them. This phenomenon is further amplified when tools based on artificial intelligence techniques are deployed. Those involved in the project also highlight the significant shift taking place with regard to digital evidence: the transition from a world where technical traces were scarce to one of abundance, or even an excess, of digital traces. This vast volume necessitates the use of filtering and selection procedures, relying on programmes that are managed and, increasingly, automated. The Digital Rights project raises numerous questions regarding this selection process, particularly concerning its documentation, the nature of the tools used to carry it out, and the opportunity for the parties to discuss the choices made.
Finally, Digital Rights’ recommendations emphasise the essential role of human assessment of relevance and probative value; it is only through such an assessment that the results of algorithmic processing chains can be regarded as evidence.
But the digital nature of traces carries with it another characteristic: they are alterable and falsifiable, to the extent that purely synthetic traces or evidence can be produced – evidence that is false but appears to be true. Whilst the history of forgery is as old as the history of evidence, the development and widespread adoption of considerable computing power, coupled with the accessibility of artificial intelligence tools, This new reality has a detrimental effect: the gradual erosion of the probative value of evidence previously regarded as strong, such as photographs and audio recordings, which are increasingly challenged in court as being false or tampered with. To anticipate this new reality, the German NGO Gewaltfrei in die Zukunft has launched a mobile app enabling victims of domestic violence to collect and certify the authenticity and integrity of the digital evidence they gather. As Stefanie Knaab pointed out, ‘difficulties in accessing justice are also difficulties in proving a case’. This example illustrates the risk mentioned above: digital evidence is also alterable evidence, the probative value of which may tend to weaken without a sufficiently robust chain of certification.
The circulation of digital evidence
The Digital Rights project has highlighted the discrepancy between the legal sphere (‘legal map’) and the technological sphere (‘technical map’). The legal landscape is territorial in nature; it is organised around the sovereignty of states as expressed in national laws. Instruments of judicial cooperation make it possible to incorporate evidence obtained in one legal landscape into another.
Conversely, the digital landscape makes it possible to offer global services with no real connection to a specific territory. With the cloud, the data itself is no longer localised and may have no real connection to the location of the defendants or the place where the offence was committed. It may even sometimes be entirely virtualised, making it impossible to predict its location at any given moment, and therefore to seize it. This evidence is often held by entities legally established in another country. Consequently, the location where the data is stored no longer determines how it is accessed.
In light of this reality, the EU eEvidence Regulation[2] appears to illustrate a transformation in the relationship between digital infrastructure and national and European sovereignties. This legislation creates new legal instruments for cooperation: European preservation orders and the European data production order. With a precisely defined scope of application, these instruments may only be invoked in criminal proceedings and for the enforcement of custodial sentences of at least four months. The entities subject to these measures are solely providers offering electronic communications services, domain name registration services, user-to-user communications services and data storage services within the EU.
The regulation marks a major shift in the administration of digital evidence and, more generally, of documentary evidence. From now on, within the framework set out by this text, in order to obtain digital records, a competent judicial authority may contact directly the private entity holding data relevant to its investigations, even if that entity is located outside its jurisdiction. If the entity to which the request is made is based outside the European Union but offers its services within it, it must have appointed a representative in an EU Member State. It should be noted, however, that this new regulation does not call into question any of the existing instruments of cooperation within the Union.
This new regulation therefore proposes a new balance between the legal sphere and the technological sphere. It brings about a new distribution of mutual trust at the heart of the European cooperation framework, through three institutional shifts. The first is the shift to enforcement by the supplier, who is contacted directly. The second is based on a redistribution of controls, whereby the service provider becomes a fully-fledged participant, and the country of enforcement may carry out compulsory enforcement in the event of a refusal by the provider and, if notified, may carry out a check prior to the handover of the data. Finally, it is for the courts of the issuing country to rule on any conflicts of law involving an obligation arising under the applicable law of a third country, and this is a genuine innovation.
The Digital Rights project makes several observations: the European text defines roles (‘issuing authority’, ‘validating authority’, etc.) and each country designates the corresponding bodies within its legal system. This necessary adaptation places a great deal of responsibility on Member States for the implementation of this cooperation mechanism. The eEvidence Regulation therefore harmonises the identification and liability of addressees, without, however, laying down all the rules governing the disclosure of electronic evidence.
This raises the key question: does this distributed architecture offer a level of safeguards comparable to traditional mechanisms of judicial cooperation? In other words: where are the safeguards for fundamental rights, and who is responsible for them? The issuing authority prior to transmission, the service provider if it identifies a problem, the country of execution prior to the transfer, and the national courts after disclosure to the parties.
## Towards a transformation of criminal proceedings?
In addition to the previous points regarding the formation, presentation and circulation of evidence, the interpretation of digital evidence can prove extremely difficult, given its volume and diversity. The judge must often rely on the investigators’ findings. How can the judge interpret the police report and assess its reliability?
The same applies to the exercise of the rights of the defence and to the very essence of criminal proceedings, which lies in the possibility – which must be strictly preserved – of discussing and challenging the evidence presented. In this regard, the _Soft Guidelines_ emphasise two further requirements: the early activation of the rights of the defence and effective access to digital evidence, which must not be limited to access to raw data. Gaia Ganeschi highlights the growing asymmetry brought about by the complexity of data processing and the tools used. To restore balance and ensure the possibility of judicial debate, she stresses that the parties’ ability to access the processing procedures put in place must be regarded as essential. Without this, how can the defence have the necessary tools to examine the results presented by the prosecution and produce its own analyses?
Taking a more forward-looking view, Lorena Bachmaier raised some important questions. Firstly, she highlights the risk of a shift from ‘justice with AI’ to ‘justice through AI’, particularly through the use of these technologies during the judicial stages of a trial. For example, to select priority cases or those that might be subject to settlement proceedings, or to verify the consistency of decisions with case law. We cannot but agree with her concerns, given that, whilst the exercise of human oversight is proclaimed, its effectiveness remains to be assessed. This use of AI tools in these decision-support phases raises an important question: to what extent do the choices made during the design, training, configuration and deployment of the systems determine the responses provided to the judge?
Building on Michael Rich’s work on ‘Automated Suspicion Algorithms’, Lorena Bachmaier also examines the consequences of mass data collection. Whilst the use of measures that infringe on civil liberties during criminal investigations is based on the existence of sufficiently incriminating evidence, might the availability of large volumes of digital evidence collected almost automatically not make it considerably easier to reach this threshold – sometimes even before the investigation has begun – thereby skewing the balance of the criminal proceedings? Given the inevitable occurrence of false positives, she considers ways of addressing this dual risk, for example by introducing stricter safeguards at the outset of investigations.
Ultimately, the researcher poses a key question that runs through the work of those involved in the project: are we still dealing with a traditional criminal procedure to which digital elements have been added, or are we witnessing the emergence of a new procedural model, distinct from the one that has gradually taken shape within the framework of political liberalism?
[1] Digital Rights Consortium, *Deliverable D3.4 – Soft guidelines on legislative implementation*, 2026, p. 16
[2] REGULATION (EU) 2023/1543 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 12 July 2023 on European Production Orders and European Preservation Orders concerning electronic evidence in criminal proceedings and for the purposes of the enforcement of custodial sentences imposed following criminal proceedings.
The proliferation of digital traces presents a new challenge for investigators, magistrates and forensic experts. The volumes of data collected are now so vast that analysis carried out solely by human means is often physically impossible to achieve within timeframes compatible with the demands of the investigation and the trial. The use of automated tools – and, increasingly, systems based on artificial intelligence – therefore appears not merely as a technological option, but as an operational necessity. However, this development raises the question of the reliability of evidence: the challenge is no longer merely to collect data, but also to understand and control the mechanisms that determine its selection and interpretation.
- 1
- 2
Français