What if the police could know in advance where a crime was going to occur? This possibility is becoming real with a new artificial intelligence tool developed by a team of American researchers.
Washington, D.C., 2054: mutant humans foresee crimes and enable the arrest of future perpetrators thanks to their gift of precognition. Such is the premise of Minority Report, released in 2002, a film that constantly oscillates between utopia and the ultimate nightmare of a society under total control.
Chicago, 2022: Ishanu Chattopadhyay and his team of researchers have developed an algorithm capable of predicting, one week in advance, the level of criminal activity likely to emerge in a given area—with an accuracy rate of nearly 90%.
Artificial intelligence, a new prophet?
To achieve this astonishing level of precision, the urban area was first divided into 300-meter-wide grids. The researchers then trained their algorithm on crime data from the city covering the years 2014 to 2016. The system successfully forecasted the level of crime for the following weeks. These results were consistently replicated in the seven other major U.S. cities included in the study.
Given such results, the key question now is how this type of modeling should be used. While it could serve as a preventive tool to ensure the safety of people and property, could it not also become a relentless instrument of social control, particularly by targeting certain populations?
This is the main criticism raised in the United States, where a previous algorithm had already been tested by the Chicago Police Department. Designed to identify individuals most at risk of being involved in a shooting, it produced a list in which 56% of those flagged were African Americans between the ages of 20 and 29—creating, according to some, a genuine risk of racial discrimination.
Focus on places rather than suspects.
For researcher Ishanu Chattopadhyay, however, his algorithm presents a crucial difference: its predictions concern only geographic areas, not potential suspects. Its use could therefore contribute to the implementation of a genuine security policy in certain regions, going far beyond merely assisting police officers on the ground.
The studies conducted by this Chicago-based team further highlighted that arrest rates were higher in wealthier neighborhoods, indicating more intensive police activity. This, in turn, raises the question of whether another form of discrimination might be at play.
Installed in commercial aircraft since 1960, the black box made its debut on July 6, 2022, in newly manufactured vehicles across the European Union. While some perceive it as a veritable “spy device,” this system is above all a valuable source of information in cases of accidents, as well as in criminal or delinquent acts.
Driver drowsiness alerts, reverse detection systems, intelligent speed adaptation… In just a few years, cars have become technological jewels generally welcomed by drivers. The introduction of this so-called black box, however—reminiscent of the flight recorders in aircraft cockpits—has generated far less enthusiasm, with critics judging the device overly intrusive.
Recording limited to a few data points.
Such concerns are unfounded, as this equipment is not intended to monitor the every move of drivers and passengers during a trip. Unlike the recorders installed in aircraft, it does not capture any sound or conversation. Nor can the data be used to identify the individuals on board or to assess driving behavior—for example, by tracking the frequency of acceleration or braking.
Its purpose? Quite simply, to analyze what happened in the event of an accident. To that end, the device records purely objective data such as speed, collision force, accelerator or brake pressure, the activation of safety systems (ABS), and seat belt usage. The electronic chip responsible for recording these data stores only the 30 seconds prior to the accident and the 15 seconds following it.
Access strictly reserved for authorities
There is no question of sharing these data with insurance companies, as some users fear. Only judicial authorities or research institutes will be authorized to access them in order to clarify the circumstances of accidents that often have tragic consequences, as illustrated by a case that occurred on the Paris ring road.
On February 21, 2013, following a 150 km/h police chase, Malamine Traoré, then 22 years old, crashed his Range Rover SUV into a marked BAC police vehicle, instantly killing two police officers and seriously injuring a third.
Beyond his numerous prior convictions for traffic offenses, his lack of a driver’s license, and a blood alcohol level of 1.4 g/L, the analysis of the black box installed in the vehicle demonstrated that the driver never lifted his foot from the accelerator nor attempted any evasive maneuver, even though he consistently denied deliberately crashing into the police car. He was tried for “intentional violence against public officials resulting in death without intent to kill” and was sentenced on appeal to 15 years of criminal imprisonment.There is little doubt that in the future, these automotive data recorders, along with other material evidence related to vehicles, will contribute to solving numerous criminal and delinquent cases.
Ukraine and numerous NGOs are building a case to bring Russia before the International Criminal Court for war crimes. On April 11, 2022, a team of 18 gendarmes from the Institut de Recherche Criminelle de la Gendarmerie Nationale (IRCGN, the French National Gendarmerie’s Forensic Science Institute) traveled to Ukraine to assist their counterparts in the difficult task of disaster victim identification. AFP interviewed Major General Patrick Touron, head of the National Gendarmerie’s Judicial Division.
The French gendarmes deployed to Ukraine on Monday have as their primary mission the identification of victims and the determination of their causes of death, explained Major General Patrick Touron, commander of the Gendarmerie’s Judicial Division, to AFP. “This team of eighteen gendarmes from the IRCGN is mainly composed of victim identification specialists, supported by two ballistics experts and an explosives specialist,” the general specified. Their mission has no fixed end date, although rotations have already been planned. “As of Tuesday, they will be in Kyiv, which for now is the location of their mission,” the general added.
The IRCGN is one of the few technical units in Europe capable of deploying to an operational theater within two hours.
Major General Patrick Touron
They will be “supervised by Ukrainian security forces” and will work alongside their Ukrainian counterparts. “The IRCGN is one of the few technical units in Europe capable of deploying within two hours to an operational theater, whether it be a theater of war (as in Mali, for example) or a natural disaster (such as the December 2004 tsunami),” General Touron emphasized. In addition to a mobile laboratory dedicated to DNA analysis, the team traveled with seven vehicles, including a truck carrying twelve refrigerated mortuary chambers to preserve autopsied bodies. The unit is fully autonomous, equipped with the means to generate its own electricity, among other logistical capacities.
We are there for the victims—to help families find their loved ones, to examine the bodies, and to determine the causes of death, – insisted the General, stressing that this was a joint effort between France and Ukraine.
Subsequently, he added, “an international structure will be established as other countries send teams on-site” to take part in the identifications and in the International Criminal Court’s (ICC) investigation into the “war crimes” committed in Ukraine. On Sunday, Ukrainian judicial authorities reported that 1,222 people had been killed in the Kyiv region since the beginning of the invasion, without specifying whether they were all civilians. Images of twenty bodies in civilian clothing lying in a street in Bucha, northwest of Kyiv, have circulated worldwide. Ukrainian authorities denounced this as a “war crime” committed by the Russian army, a charge immediately denied by Moscow. On Friday, President Emmanuel Macron declared that France was in the process of “gathering evidence” of “Russian war crimes” in Ukraine and announced the deployment of French gendarmes and magistrates to the country.
DNA is everywhere—in water, soil, and even in the air. Today, scientists have at their disposal analytical methods capable of capturing and sequencing this environmental DNA (eDNA).
Thanks to major advances in PCR (Polymerase Chain Reaction) and sequencing technologies, the use of environmental DNA, or eDNA, has gained considerable momentum. The most recent breakthrough in this ever-evolving field: the collection and sequencing of DNA present in the atmosphere.
Environmental DNA: a reliable and non-invasive technique.
The principle of eDNA is based on the fact that all living organisms shed DNA fragments as they move through their environment—through urine, feces, hair, scales, secretions, and more. These genetic traces accumulate in the environment: in the ocean, rivers, lakes, soil, sediments, and, as demonstrated by two studies conducted in zoological parks, even in ambient air. Such traces provide valuable information for detecting the presence or passage of species—from microorganisms such as bacteria and viruses to elusive or rare species that are difficult, if not impossible, to monitor using conventional methods. Another key advantage of eDNA is its non-invasive nature. Only a small sample of the medium inhabited by the target species is required, without any direct intervention on the organisms themselves.
From simple traces to DNA barcoding.
Once collected, water or soil samples are sent to the laboratory for analysis. The protocol, consisting of no fewer than four steps, requires the utmost rigor on the part of scientists in order to avoid the risk of contamination.
The first step involves isolating DNA molecules using molecular biology techniques (such as precipitation and centrifugation), which are now well mastered and largely automated, thereby reducing costs. In a second step, this DNA is amplified through PCR to obtain sufficient quantities for sequencing.
DNA sequencing makes it possible to determine the sequence corresponding to a given species—that is, the specific fragment of nucleotides that characterizes and differentiates it from others. These sequences are then catalogued in international databases according to the concept of molecular barcoding, developed in 2003 by Canadian zoologist and ecologist Paul Hebert. Stored in the form of barcodes and integrated into computerized systems, DNA fragments can thus be easily compared and identified in record time.
A vast field of applications.
Within just a few years, environmental DNA (eDNA) has become a particularly valued tool for biodiversity studies. By analyzing environmental samples, researchers can identify and quantify organisms present in natural habitats, monitor rare or endangered species, or, conversely, track the emergence of invasive alien species that may disrupt ecosystems. Even more striking, the analysis of a single honey sample can reveal all the plant species visited by bees, while DNA trapped in just a few grams of sediment can provide insights into which species inhabited specific regions thousands of years ago.
Most recently, this technique has made a significant contribution to the fight against the Covid-19 pandemic. In this case, viral RNA was detected in wastewater, allowing researchers to quantify the presence of the virus and, consequently, to better anticipate the progression of the epidemic. In the future, this tool could also be used to identify the emergence of new variants at an early stage—pending further applications, equally fascinating, in many other fields.
The path that leads from the criminal act to its perpetrator is what we call evidence. In various forms, it is the daily concern of investigators, prosecutors, and judges. Its essentially retrospective nature makes it difficult, unpredictable, and uncertain. One cannot simply turn back time. Years of experience at the Assize Court have taught me that the path separating the most spectacular piece of evidence from a declaration of guilt is a mysterious one: just when you think you have it, it slips away; it may seem obscure, then suddenly becomes compelling. “You who enter here, beware: absolute proof does not exist” could well be the warning engraved above the doors of forensic laboratories. And rightly so.
When it’s too good to be true
He had just killed the elderly lover who had refused to comply with his financial demands when, seized by remorse, he alerted the police in the hope that they might provide the medical care which, he thought, could save him. Without revealing his identity, he fled before the emergency services arrived. Arrested shortly thereafter and confronted with the recording of his call, he admitted to the facts. However, part of the procedure containing his confession was annulled. When the investigation resumed, he chose to change his line of defense and denied the facts, a position he maintained until trial. The presiding judge allowed him ample time during his personality examination so that everyone could become familiar with the sound of his voice. Then came the review of the facts and the playing of his recorded message by the police.
During the adjournment—which is a time for relaxation and informal exchanges—all the judges and jurors recognized the accused’s voice without hesitation. The case seemed settled. Except that…
When the session resumed, and to drive the point home, the prosecution requested that the incriminating message be played again, which was done. At the next adjournment, however, one of the jurors began to have doubts. At the request of the civil party, the message was played again, but after the following recess, three jurors were questioning it. Needless to say, when asked one last time to order the message to be replayed, the presiding judge, now wise with experience, refused. The accused, whose retracted confession was known only to the professionals, might well have been acquitted if the playback of his call continued indefinitely.
The more conclusive a piece of evidence appears, the more cautious one should be
His very distinctive haircut at the time made him recognizable among a thousand: shaved all around the head, leaving only an oval brush of thick black hair on his high, slender forehead. None of these details escaped the video surveillance cameras in the underground car park he had entered barefaced to commit his crime. He was acquitted.
What better proof than to see the victim herself designate her killer with her own blood—moreover, by making a spelling mistake she habitually committed? And yet, the conviction of the accused did not prevent the movement which, invoking a miscarriage of justice, ultimately led to his pardon.
“What are the chances of a mistake in the identification of this DNA profile?” experts are frequently asked. “About one in a billion,” they reply. Until the day it becomes apparent—as I once witnessed—that a material error in an expertise hastily dictated had falsely designated the accused.
Examples of this kind abound, and one may draw a first lesson from them: the more conclusive a piece of evidence appears, the more cautious one should be.
To make progress over and over again
To make progress over and over again
None of this should discourage technicians and experts from constantly striving to refine investigative techniques, and it must be acknowledged that since the discovery of dactyloscopy, the progress made in identifying criminals or exonerating suspects has been remarkable.
Scientific research is continuous, and we can hardly imagine what future advances will further shed light on the truth.
Yet, however far progress may take us, there remains a discontinuity that will always separate the administration of evidence from the declaration of guilt, and which will sometimes frustrate even the finest investigators: that of the work of reason.
Without reproducing in full the address that the presiding judge delivers to the jurors at the end of the trial, before the court withdraws to the deliberation chamber, it is worth quoting a passage contained in what is arguably the most beautiful article of the French Penal Code—Article 353, which establishes the freedom of evidence and the principle of intimate conviction: “The law requires the judges to seek in their conscience what impression the evidence presented against the accused and the arguments for his defense have made upon their reason.” However perfect the evidence may be, its demonstration will always demand an exercise of reason, without which no finding of guilt can be pronounced. And there is nothing more uncertain than reason, even if collegiality greatly reduces its unpredictability.
The challenges of deliberation
It is, for example, particularly difficult to rule on homicidal intent. What does it mean to “intentionally cause death”? Must one probe the exact thoughts of the accused at the precise moment of an act to which he may not even have given a thought an instant earlier, and which he would regret as soon as it was committed? Did he truly intend to cause death? Quite clearly, that is impossible. It must therefore be determined whether death was the logically foreseeable material consequence of an act committed by a conscious and lucid individual.
Similarly, DNA traces found on the complainant’s underwear in a rape case, however categorical they may be, will never dispense with the need to question consent; nor will defense injuries observed on her wrists relieve the court from inquiring into the origins and circumstances of the struggle.
Not to forget—and this may be the essential point—that a trial, where doubt benefits the accused, is the singular exercise that requires judges to determine what they are certain of, in relation to facts to which they were not witnesses.
Back to the basics
Investigators and experts are, of course, fully aware of all this. But to this retrospective approach, which is their daily lot, there succeeds another—this time prospective—which is precisely the mission of that interface known as the Public Prosecutor’s Office. Its role is to assess, with all possible caution given the uncertainty that characterizes the judges’ task, and to guarantee the quality of the evidence that the prosecuting authority will submit to them.
In this uncertain—and, to be truthful, rather vertiginous—procedural chain which, beginning with the initial finding of the crime, must lead with sufficient certainty to the identification of the criminal and the punishment of the offense, it is essential that each actor be fully conscious of his or her own role and of the place he or she occupies.
Conflict zones often provide an opportunity to test and deploy new technologies. Such is the case with facial recognition, which Ukraine is set to use for identification purposes. For better or for worse?
In his novel 1984, George Orwell imagined the character “Big Brother,” whose watchful eye constantly surveils the population. In 2022, Big Brother has taken the form of facial recognition, which is gradually spreading within our societies, despite its controversial reputation.
In the context of the war between Russia and Ukraine, this technology—made available to Ukrainian authorities by the start-up Clearview AI—is intended to identify refugees at checkpoints, to recognize individuals killed in combat, and to detect Russian agents attempting to infiltrate. Clearview AI’s search engine relies on a database of more than ten billion images, sourced in particular from social media.
Between biometrics and Artificial Intelligence.
Facial recognition analyzes the distinct characteristics of a face by combining several technologies: biometrics, Artificial Intelligence, and 2D or 3D mapping. In practice, a face is first isolated from a photograph or video and its specific features examined (such as the distance between the eyes, the size and position of the ears, or the shape of the lips). Facial recognition software can process up to 80 of these characteristics, also known as nodal points. These data are then used to generate a numerical code or “facial imprint” unique to each individual, much like fingerprints. This imprint can subsequently be compared against a database containing millions of other similarly mapped faces. Thanks to artificial neural networks modeled on the human brain, deep learning has enabled algorithms to acquire the ability to recognize human faces and, in principle, to match a proposed imprint with the corresponding photograph without error.
An exceptional tool in the field of security.
Even if it is not always apparent, the use of facial recognition technology is steadily expanding. Authorities are increasingly employing it, particularly for surveillance in airports and at border crossings.
In the United States, where no law regulates the collection and storage of personal data, the FBI already maintains a database of 650 million images, sourced from airports and social media. In many U.S. cities, law enforcement officers are also equipped with body cameras capable of real-time facial recognition. A simple photograph of a driver or a suspect can be cross-checked against available databases to determine whether that individual is on record.
In France, police forces have access to a database incorporating facial recognition known as TAJ (Traitement des Antécédents Judiciaires). It consolidates information drawn from investigative and intervention reports and currently contains more than 8 million photographs. Access to this database is legally regulated and restricted to criminal investigations, inquiries into misdemeanors, and the search for missing persons. In addition, the PARAFE facial recognition system has been deployed in a number of strategically sensitive train stations and airports.This technology is undeniably a valuable tool for securing major events, tracking fugitives, or identifying dangerous individuals. However, France remains cautious about its use. While the CNIL (Commission Nationale de l’Informatique et des Libertés) authorized an experimental trial in 2019 during the Nice Carnival, the organizers of the 2024 Olympic Games appear to have definitively ruled out its use as a security measure.
Privacy under surveillance?
Such caution is far from whimsical. Beyond the obvious difficulties in regulating its use and preventing abuses in conflict zones, facial recognition immediately raises a fundamental question: how can individuals protect their privacy and personal data if their faces risk being incorporated into such software without their knowledge?
Indeed, many companies and online platforms have already integrated facial recognition into their technologies. Apple, for instance, uses it to unlock smartphones, while Twitter, Facebook, and Google have also implemented similar systems. This has already enabled Clearview AI—the very company now equipping Ukraine—to harvest several billion photographs, videos, and personal data. The consequences of such practices remain difficult to fully assess.
Chimerism has long since left the realm of mythology and entered that of science. Today, it is an undeniable human reality, particularly through organ and bone marrow transplants. One human being, one DNA—that was the rule that had guided scientists for centuries. With advances in medical research, however, this fundamental principle has been completely overturned, as illustrated by the case of Chris Long.
When DNA blurs the trail.
This American patient, suffering from acute myeloid leukemia, received a bone marrow transplant from an anonymous European donor. This procedure allows the donor’s hematopoietic cells (blood-forming cells/blood cells) to replace the recipient’s non-functional blood cells. It is therefore not surprising to find foreign DNA in the blood of the transplanted patient!
In the Chris Long case, however, the surprise was total when analyses carried out four years after the transplant revealed that his seminal fluid contained only the donor’s DNA—in this case, that of a young German. Other parts of his body, such as his cheeks, tongue, and lips, contained both DNAs, whereas his hair contained only his own. Specialists were thus confronted with a case of chimerism, a condition in which a person carries two different genetic profiles.
A natural anomaly.
The phenomenon is extremely rare. Only about a hundred cases of chimerism have been recorded worldwide. In Chris Long’s case, this anomaly arose in the context of a medical intervention, but it can also occur naturally when an incident disrupts the developmental process. In a dizygotic twin pregnancy—meaning two embryos from separate fertilizations—it may happen that the two eggs fuse at a very early stage of development. The surviving embryo is then formed from cells originating from both eggs and therefore carries two distinct DNAs.
Most often, chimerism affects specific organs, body parts, or certain cell types such as those of the blood group. A striking example occurred in 2015, when another American underwent DNA testing that appeared to prove his son was also his nephew. The mystery was quickly solved when it was discovered that the father actually carried two DNAs—his own and that of a twin brother who had never developed in utero.
A textbook case for forensic science.
Although chimerism is rarely identified (it is often detected only through prenatal diagnosis), this is largely because it has no impact on the health of the person concerned. Nevertheless, it is of growing interest to specialists in forensic medicine and forensic police. In just a few decades, DNA has become the cornerstone of investigations aimed at solving numerous criminal cases.Chimerism, however, could change the rules. If, in the context of a sexual assault, DNA recovered from the crime scene belonged to a person who had undergone a bone marrow transplant, could this not lead to the wrongful implication of an innocent donor? Will it now be necessary to collect DNA from multiple parts of the body to avoid identification errors with potentially dramatic consequences? These are pressing questions to resolve—and a new textbook challenge that forensic procedures must now take into account.
Each year, the forensic units of the National Police and the National Gendarmerie detect and recover several hundred thousand fingerprint, palm-print and even footprint impressions on a wide range of surfaces, linked to hundreds of thousands of delictual or criminal offences (burglaries, vehicle thefts, drug trafficking, armed robberies, rapes, homicides, terrorist attacks, etc.).
– text translation in order of appearance: Friction ridge impressions Crime scene – do not cross Ridge ending, lake, short ridge / fragment ; suitable print First responder, victim, witness, janitor, relative of the victim ; suspect No match Loading 6,5 millions (in 2021) ; Automated Fingerprint Database (FAED) (french equivalent of AFIS) data transfer into the database ; retention period 15 to 25 years ; fingerprint hit —
Since the Francisca Rojas case in Argentina in 1892 (1) and the identification of Henri-Léon Scheffer in France in 1902 by Alphonse Bertillon (2), this form of physical evidence has made it possible to identify offenders through the latent prints they leave behind at crime scenes.
International practice in fingerprint identification is not standardized. In France, in order to link a latent print recovered from a crime scene to a friction ridge impression taken from a suspect, forensic experts rely on two approaches:
• The 12-point numerical standard
• The probabilistic or holistic approach
For nearly a century, the binary view of fingerprint identification based on the “12-point standard” has gradually given way to a continuum of possible conclusions. Depending not only on the number but also on the quality of the minutiae present in the questioned print, experts use a numerical and verbal scale of comparative assessment that may point more or less strongly toward identification or exclusion.
At present, this probabilistic approach is applied only in complex fingerprint examinations. The 12-point numerical standard continues to be used routinely in France—see our video on the recovery and comparison of friction ridge impressions.
Amendment to Article 55-1 of the French Code of Criminal Procedure
In the course of a flagrante delicto investigation, Article 55-1 of the French Code of Criminal Procedure expressly provides that, during the judicial inquiry, samples may be taken from any person concerned by the procedure. Such samples may prove necessary to carry out technical and scientific examinations for comparison with traces and evidence collected as part of the investigation. Under what conditions may these samples be taken?
What does Article 55-1 of the CCP provide?
The judicial police officer may, personally or under his or her supervision, collect external samples from any person likely to provide information on the facts in question, or from any person against whom there are one or more plausible reasons to suspect that they have committed or attempted to commit the offence. These samples are necessary to carry out technical and scientific examinations for comparison with traces and evidence collected for the purposes of the investigation.
The officer also carries out, or has carried out under his or her supervision, the collection of identifying information, in particular the taking of fingerprints, palm prints or photographs necessary for the feeding and consultation of police databases, in accordance with the rules applicable to each of these files.
Refusal by a person against whom there are one or more plausible reasons to suspect that they have committed or attempted to commit an offence to submit to the sampling operations mentioned in the first and second paragraphs, as ordered by the judicial police officer, is punishable by one year of imprisonment and a €15,000 fine.
In decision no. 2003-467 DC, the Constitutional Council held, with respect to Article 30 of the Law of 18 March 2003 on Internal Security, ‘that the expression “external sampling” refers to a sampling procedure that does not involve any internal bodily intervention; that it therefore entails no painful, intrusive or degrading methods affecting the dignity of the persons concerned; that, accordingly, the claim alleging a violation of the inviolability of the human body is unfounded; that external sampling does not, moreover, adversely affect the individual liberty of the person concerned’ (para. 55).
Fingerprinting under coercion:
Article 30 of Law no. 2022-52 of 24 January 2022 on Criminal Liability and Internal Security supplements Article 55-1 of the French Code of Criminal Procedure with the following paragraph:
« Without prejudice to the application of the third paragraph, where the taking of fingerprints, palm prints or a photograph is the sole means of identifying a person interviewed under Articles 61-1 or 62-2 for a felony or a misdemeanour punishable by at least three years’ imprisonment, and who refuses to provide proof of identity or supplies manifestly false identity information, this operation may be carried out without that person’s consent, with written authorisation from the public prosecutor, upon a reasoned request by the judicial police officer. The judicial police officer, or under his or her supervision a police officer, shall resort to coercion only to the extent strictly necessary and in a proportionate manner, taking into account, where appropriate, the vulnerability of the person. A report shall be drawn up of this operation, stating the reasons why it constituted the sole means of identifying the person, as well as the date and time when it was carried out. The report shall be transmitted to the public prosecutor, with a copy provided to the person concerned. »
Finally, a French website that aims to promote not only the reliability of results through the acceptance of questioning the knowledge and truths proclaimed by “experts” (which aligns rather well with our distinctly French critical mindset), but also the emergence of a new discipline: “forensic” science, which deals with inferences drawn from the study of effects in order to approach their underlying causes. On this site, we will therefore not be subjected to dogmas, technological quarrels, or axioms, but rather be encouraged to awaken a “forensic” way of thinking that is open to society, to critique, and firmly modern and universal in its technical application.
To assist the reader of this site, it is essential to define the terminology employed by the “experts” working towards the pursuit of truth in criminal trials within the judicial system. Just as law, the Navy, or medicine have their own specialized vocabulary, so too do the sciences, and the use and precision of this vocabulary are necessary to understand their results or demonstrations. The absence of bias in the reporting of findings and the linguistic clarity of conclusions largely contribute to the acceptance of judicial decisions or of the “right to punish” that is vested in Justice in our societies.
It is therefore indispensable to distinguish “criminology,” which concerns itself with the social phenomenon of crime, from “criminalistics,” which is the use of scientific methods, belonging to the exact sciences, to study or evaluate traces found at crime scenes. The Code of Criminal Procedure, in its Article D.7, refers to “technical and scientific police operations,” in order not to differentiate between the detection, collection, preservation, or exploitation of traces. These specialties will not be addressed directly in this article. Instead, we will consider the field of “forensic sciences,” representing the practical application of all the “exact” sciences to the pursuit of truth in criminal trials, sciences whose role is to shed light for the judge in the decision-making process. To illustrate this point, I would say that if the position of celestial bodies were relevant to solving a crime, forensic sciences would include astrophysics, but not astrology.
Naturally, this leads us to the emergence of “Forensic” science in the singular, to demonstrate that, as a result of numerous theoretical contributions regarding the study of effects to determine causes, the reasoning and techniques employed in approaching the crime scene have now given rise to a distinct science in its own right: “forensic,” a discipline open to history, economics, and innovation, but also to evolution and intrinsic critique, no longer limited to the mere application of another science (physics, chemistry, genetics, etc.) to the judicial domain of trace evidence examination. As with other sciences, peer-reviewed journals dedicated specifically to forensic science have thus emerged.
Forensic Science Institute of the French Gendarmerie – IRCGN
From the moment justice sought to uncover the truth through demonstration rather than relying on divine judgment (ordeals), it turned to knowledgeable individuals in relevant fields for guidance. Within the inquisitorial system, as early as the fifteenth century, midwives, surgeons, and physicians were called upon by the courts to determine the nature of a crime, cases of infanticide, to examine a body in order to establish the time of death, the type of injuries, their lethality, or to identify the weapon used. Thus, given the diversity of crimes, all trades quickly became “required,” such as notaries for forged documents, goldsmiths for counterfeit jewelry, or carpenters for break-ins, until the development of specialized sciences in the late nineteenth century brought increasingly advanced expertise—chemistry, toxicology, physics, and even mathematics, as exemplified during the Dreyfus trial. Today, experts organized into professional associations assist magistrates in nearly every country in the world.
The more justice seeks precision in its decision-making, the more it turns to experts to shed light on its judgments and establish the moral or material causality of offenses. France occupies a special place within the global forensic community, due to the theorization, in the early twentieth century, by Dr. Edmond LOCARD[1], of this specific approach to the crime scene, through the formulation of a fundamental principle: “every contact leaves a trace. It is impossible for a criminal to act, especially considering the intensity of a crime, without leaving numerous marks of his presence. Sometimes the offender leaves behind traces of his activity at the scene, while at other times, through a reverse action, he carries away on his body or clothing indications of his presence or his actions.” This clear statement underlines the necessity of trace evidence detection by investigators at crime scenes in order to substantiate the offense and identify its perpetrator.
A second principle essential to forensic sciences was articulated by Paul Leland KIRK (1902–1970): “Every object in our universe is unique. Two objects of common origin can be compared, and individualization can be established if the objects possess sufficient quality to permit the observation of their individuality.”
When these two fundamental principles are combined, one arrives at the scientific demonstration of the individualization of traces and, potentially, of the offender or the location from the traces recovered at a crime scene. Forensic experts constantly strive to identify such individualizing traces, which today are best exemplified by the detection of biological traces left by a perpetrator at a crime scene, enabling the identification of their unique source.
It can thus be concluded that criminalistics seeks to detect and analyze traces and evidence recovered from crime scenes on the basis that the study of effects allows one to infer causes, by applying the first principle—that every contact leaves a trace—and the second—that every trace is individualizing—together with the fact that no two random phenomena ever leave exactly the same traces. Nevertheless, this requires particular caution on the part of those tasked with interpreting results, since the traces discovered by investigators—and those specifically sought[2]—do not reflect all the traces actually present, nor therefore all the effects corresponding to all possible causes.
Signal, Image, and Speech Department (SIP) – Credits: Forenseek
The holistic approach to trace analysis drives innovation, leading industry to continually develop techniques for improved detection (such as the crimescope[3] in the 1980s), better collection of highly informative evidence like genetic traces (swabs in the 2000s), the recovery of invisible digital evidence (radio wave sensors/IoT devices in the 2010s), and, more recently, the development of odor sensors capable of detecting olfactory signatures[4] potentially left behind by an offender wearing a mask and gloves. All these new capabilities are now available to forensic police units worldwide. Forensic science laboratories, meanwhile, employ analytical equipment that is increasingly sensitive, specific, and rapid—producing results that are ever more complex to interpret, and requiring the expertise of highly skilled professionals.
When all the data obtained from the exploitation of these different types of evidence are combined, it becomes indispensable to rely on artificial intelligence tools or data-mining methods, which are now prerequisites for understanding, contextualizing, and exploiting the full range of information derived from a criminal case. Whereas in the 1980s a typical case file comprised around 1,000 pages (readable by a single person), today the investigation involves the processing of gigabytes of data in the form of texts, images, recordings, connection logs, audio files, and a wide variety of highly specialized analytical results. Furthermore, each item of evidence carries a specific weight in the demonstration (Bayesian interpretation), which may vary depending on the context of its discovery. For instance, a DNA trace recovered from a cigarette butt (which could have been transported) does not carry the same probative value as the same DNA trace found on a knife embedded in a victim’s back. In forensic science, this relative probative value is formalized using Bayes’ theorem, which allows the probability of one event to be determined from another event that has occurred, provided the two events are interdependent. Here again we find the pursuit of inference through the study of effects to understand their causes.
Only the use of tools capable of automatically creating relational links within a case file, of computing Bayesian networks to assess the weight of each piece of evidence, of annotators able to contextualize words within sentences or detect temporal or geographical inconsistencies between statements, or of systems able to retrieve a specific image from video recordings, will make it possible to fully comprehend the case, to reconstruct events—in short, to “judge.” The development of tools with fully controlled algorithms, free of the cognitive biases of their designers, is currently under evaluation in forensic science laboratories.
Department of Anthropology and Hemato-Morphology (ANH) – Credits: Forenseek
The capture of all traces present, in every form that technology allows, and their placement within a digital space reconstructing the crime scene, already provides us today with a tool enabling the visualization of such scenes (the digital twin). Yet, far beyond a simple animated reconstruction in augmented reality, the advent of the “metaverse,” integrating all the previously described digital tools, heralds a paradigm shift. A « metaverse » capable of reproducing natural laws (all physical phenomena of the real universe)—the trajectory of a projectile, the fall of a body, the explosion of a bomb and the projection of fragments, even the reconstruction of blood spatter—will enable us to replay scenarios, generate hypotheses, and test them in a theoretical effect/cause feedback loop that can propose alternative sequences of events, redirect the search toward specific traces, or reveal inconsistencies in the discovery of certain evidence. The integration of probabilistic laws into these metaverses will provide investigators and magistrates with a scoring system ranking one hypothesis as more probable than another in light of the traces recovered. The ability to process enormous quantities of data, combined with artificial intelligence algorithms capable of proposing and replaying hypotheses and scenarios, simulating them, and then analyzing their effects to compare and progressively (mathematically) align them with traces found at the crime scene, represents the ultimate step in understanding an event. This should allow the magistrate to come as close as possible to the reality of what occurred.Forensic science, which seeks to reconstruct the past from thousands of data points gathered in the present, opens a field of critical analysis that helps to nourish and shape the overall thinking of our society. Market economies, urbanization, environmental issues, and pandemics raise the same questions of why a situation arises, in order to better adapt to it. Forensic science, in its pursuit of judicial truth in criminal trials, has developed theoretical foundations and tools to address this why and propose a how, thanks to the instruments designed by investigators and forensic scientists.
Références
[1] Edmond Locard (1877-1966) is the author of a Treatise on Forensic Science (Traité de police scientifique) in seven volumes. This work proposes a methodology for this new science and is still used today as a foundation for all forensic science laboratories worldwide. This treatise includes a study of criminal investigation, proof of identity, fingerprints, and the examination of written documents, among other subjects.
[2] “The eye sees in things only what it looks at, and it only looks at what is already in the mind.” Motto attributed by Lacassagne to Bertillon (Niceforo, 1907)
[3] Crimescope: Multispectral lighting allowing the selection of specific wavelengths for certain traces (fiber, blood, etc.) and thus making them detectable in an environment where they were not visible to the naked eye.
