France

• Introduction and background
• Part 1: Legal frameworks
• Part 2: Policies and procedures
• Part 3: Comparative indicators
• Summary and recommendations
• References and data sources

Introduction

Abstract

Certain features of the French system constitute relatively good practice when looked at internationally, but more needs to be done, including on data collection, comparability and publication, and learning lessons. It has been possible to construct most of the indicators regarding numbers of deaths and injuries, albeit sometimes using unofficial sources. It would be helpful to have public and official data on the number of officers killed, the number of arrests and the number of firearms seized.

For the year 2019, the number of civilians killed by on-duty public security agents for every 100,000 inhabitants was 10, the number of civilians killed by on-duty public security agents for every 1,000 public security agents was 0.04, and the ratio of the number of civilians killed by gunshot by on-duty public security agents to public security agents killed by gunshot in homicides while on duty was 10/0.

Very little data is provided on the civilians killed by law enforcement officials. Gender should be recorded more systematically. It is not legal to produce ethno-racial statistics in France. However, in order to be able to measure inequalities in relation to policing and law enforcement, it would be desirable to have more information on the nationality of the victim and their parents, whenever possible.

The authorities seem reluctant to produce any analysis of the use of force (lethal or less lethal) in policing and law enforcement activities. Similarly, the 2017 law on the rules governing the use of firearms, its implications in practice and its compatibility with human rights law do not seem to be questioned since the law came into force.

Background

Law enforcement agencies in France consist of two main bodies: the Police Nationale (the national police, referred to here as the police) and the Gendarmerie Nationale (the national gendarmerie, referred to here as the gendarmerie). The police is a specialized civil body whereas the gendarmerie is a military body and part of the army. There are currently approximately 150,000 police officers and 100,000 gendarmes in France. According to a long-standing practice, the police is primarily deployed in urban areas (cities with more than 20,000 inhabitants), while the gendarmerie mainly operates in rural areas. In practical terms, this means that the gendarmerie covers 95 per cent of the French national territory and 50 per cent of the population. Both bodies have been under the authority of the Ministry of the Interior since 2009. In addition to these two main law enforcement organizations in France is the municipal police (approximately 23,000 officers), which operates under the control of the mayor of each municipality and consists of agents employed by local administrative bodies (communes). Only the national police and gendarmerie are covered in this report. These two national law enforcement agencies are overseen and held accountable by two inspectorates, the General Inspectorate of the National Police (Inspection Générale de la Police Nationale or IGPN) and the General Inspectorate of the National Gendarmerie (Inspection Générale de la Gendarmerie Nationale or IGGN). The police and gendarmerie can also be subject to investigations by the Defender of Rights (similar to an ombudsman).

Part 1: Legal Frameworks

Global treaties

1966 International Covenant on Civil and Political Rights (ICCPR)		State party
ICCPR Optional Protocol 1		State party
1984 Convention Against Torture (CAT)		State party
CAT committee competent to receive individual complaints?		Yes
CAT Optional Protocol 1		State party

Regional treaties

1950 European Convention on Human Rights

State party

National legal provisions

Relevant constitutional provisions or general laws

France has a written constitution but this document does not contain provisions applicable to the use of force by LEOs.

The use of force is partly covered by the general provisions of Article 122 of the Penal Code, including:

• Article 122-4 which excludes criminal liability for acts authorised by law or ordered by a legitimate authority;
• Article 122-5 and 122-6 which exclude criminal liability for proportionate acts carried out for the legitimate defence of oneself or another against an unjustified attack, or to prevent a property offence;
• Article 122-7 which excludes criminal liability for proportionate acts carried out in response to an actual or imminent danger where it is necessary to act to protect a person or property.

For more details, see the summary at https://www.policinglaw.info/country/france which includes Law no. 2017-258 (28 Feb 2017) amending Article L435-1 of the Internal Security Code. This law also repealed Article 122-4-1 of the Penal Code (see next).

Relevant specific national legislation

The use of force is covered by the Code of Ethics of the Police and the Gendarmerie in France. Article L434-1 of the Internal Security Code states that the Code of Ethics is established by a decree issued by the Conseil d’État. The Code of Ethics is included in Book IV, Title III, Chapter 4 of the regulatory part of the Internal Security Code. It is also available on the Interior Ministry website.

The use of force by the gendarmerie is authorised by Article L2338-3 of the Defence Code, which refers to Article 435-1 of the Internal Security Code.

The use of force by the police and gendarmerie is now specifically covered by Article 435-1 of the Internal Security Code, as amended by Law no. 2017-258 (28 Feb 2017), to bring the use of firearms by police officers into line with those for gendarmes. Previously, police officers were subject to the Penal Code and had to prove self-defence like any other citizen, whereas gendarmes could neutralize an individual who was trying to escape from them and risked harming third parties in the process. It is worth noting that the draft bill’s impact study also refers to the Guerdner v. France judgment handed down on April 17, 2014 (cited below) to justify the compliance of French legislation with Article 2 ECHR¹.

Following the 2017 law there has been an increase of police shootings involving firearms, in general² and against moving vehicles³, but also an increase of fatal police shootings against moving vehicles⁴. A number of critics have argued that the law introduced confusing notions and questionable interpretation of the use of firearms. As Catherine Tzutzuiano, legal expert, argues, the legislator has legally enshrined an act of ‘anticipatory defense’. Although the rules of engagement remain close to the framework of self-defence, they create considerable confusion⁵. “This has created a risk of increased use of weapons, and a legal risk for police officers, because they remain criminally liable”⁶.

As of March 1, 2017, both the Directorate-General of the National Police(Direction générale de la Police nationale or DGPN) and the Directorate-General of the National Gendarmerie (Direction générale de la Gendarmerie nationale or DGGN) have issued instructions concerning the application of the law. The DGGN recalls the importance of immediate danger in its rules governing the use of weapons⁷. However, for the DGPN “unlike self-defence, the condition of simultaneity is softened [ … ] It is not required that the individual immediately and directly threaten [the police officer] or others in order to use the weapon against him. But the police officer must, at the moment he uses his weapon, have real and objective reasons to believe that this individual is dangerous, i.e. likely to harm his life or physical integrity or those of others”⁸.

Flagrant Déni, a French militant media outlet on police impunity, has revealed that the police abrogated the 2017 instruction in 2021⁹. A new instruction on individual or service weapons was issued on May 26, 2021. The document is not public and the national police force did not wish to share it with Flagrant Déni. However, the media was able to consult it and considers that the DGPN maintains the confusion in the rules on firearm use.

Relevant national regulations

In 2020, the French Ministry of the Interior published the national policing plan for maintaining public order (schéma national du maintien de l’ordre or SNMO), a set of guidelines for all law enforcement agencies concerning the management of demonstrations. Following the repeated questioning of policing strategies during the Yellow Vest movement (a national protest movement in France that began in 2018), the Minister of the Interior formed a “working seminar” aimed at adapting and updating the management of public order in France, which ultimately led to the SNMO. The SNMO endorses practices that have been tested in the field in recent years, including assigning units trained to maintain order to static protection missions, while the non-specialized units are tasked with intervening in the body of the demonstration; and legitimizing the use of weapons developed to tackle urban violence, such as kinetic impact projectile (KIP) launchers¹⁰ (known in French as lanceurs de balle de defense or ‘defensive ball launchers’, which fire hard foam rubber balls).

About less lethal weapons in France

Less lethal weapons were introduced in France in the 1990s for specific police units, and have been generalized since 2004¹¹. Over the years, there have been a number of different models of stun grenades (grenades à effet de choc), thrown by hand or by propulsion, ‘sting ball’ grenades (grenades de désencerclement), which project small rubber balls designed to strike shins and calves, and KIP launchers. All these weapons are designated as war materiel by French legislation.

France is one of the few European countries —along with Spain, Poland and Portugal to a lesser extent,— to make extensive use of KIP launchers¹². According to figures communicated by the Ministry of the Interior in 2020, 2448 people were injured and 19,071 shots were fired with KIP launchers, 1,428 tear-gas grenades were fired, and 5,420 sting ball grenades were used during the Yellow Vest movement¹³.

Taser has been authorized for use by the police since 2007 and is being used more and more frequently by the police. The IGPN reports that in 2022, 2995 uses of tasers were recorded in the police force¹⁴. This is twice as many as 6 years ago (1,403 shots), and 5 times as many as in 2014 (522). The National Gendarmerie does not provide similar data.

The Maintien de l’ordre website brings together all the technical details on the less lethal armament of French law enforcement forces¹⁵. The Violences policières website lists, among others, the victims of less lethal weapons¹⁶. Both websites are in French only.

Legal interpretation and application

UN or other international body decisions or advisory opinions

In 2015, in its Concluding Observations on France, the Human Rights Committee expressed its concern about “allegations of ill-treatment, the excessive use of force and the disproportionate use of non-lethal weapons, especially during arrests, forced evictions and law enforcement operations.”

The Committee was “further concerned about continued racial profiling and allegations of police harassment, verbal abuse and abuse of power against migrants and asylum seekers in Calais”¹⁷.

In 2016, the Committee against Torture also expressed its concern about “the allegations of excessive use of force by the police and the gendarmerie, which has in some instances led to serious injuries or death”. It was also concerned by:

1. reports of problems faced by victims in filing complaints;
2. the lack of statistical data on complaints filed that would allow for comparisons with inquires launched and cases prosecuted;
3. the lack of detailed information on related convictions of police and gendarmerie officers and the sentences handed down; and
4. reports of high numbers of cases being dismissed or discontinued, light administrative sanctions being imposed that are not proportionate to the seriousness of the actions, and a very small number of court-ordered penalties being imposed upon police and gendarmerie officers¹⁸.

The Committee was further concerned “about the allegations of violence being used against asylum seekers and migrants, and about their situation in Calais and the surrounding area”.

In 2019, after urging France to halt the use of kinetic impact projectiles, the Council of Europe Commissioner for Human Rights, Dunja Mijatović, invited the French authorities to show more respect for human rights during operations aimed at maintaining public order and to refrain from introducing excessive restrictions to freedom of peaceful assembly through the bill on strengthening and guaranteeing public order at demonstrations¹⁹.

In 2022, the European Commission against Racism and Intolerance (ECRI) called on France to improve the recording of police identity checks²⁰.

In 2023, the UN Committee on the Elimination of Racial Discrimination, CERD, adopted a statement on the situation in France, expressing its concerns and making a number of recommendations to the State regarding racial profiling and excessive use of force by law enforcement officials²¹.

In 2023, the Monitoring Committee of the Council of Europe’s Parliamentary Assembly, which is responsible for verifying fulfilment of obligations assumed by member states of the Council of Europe, “considered many longstanding issues concerning the functioning of democratic institutions, the rule of law and human rights in France”. Among other statements, “the Assembly is particularly alarmed by the high number of people injured during demonstrations, especially the number of injuries with serious long-term consequences. In this connection, it regrets the fact that the official statistics do not provide a clear picture of the number of people injured or killed by law enforcement officers during demonstrations or the number of such officers sanctioned or having received criminal convictions for unlawful acts of violence committed during the demonstrations. Having such statistics would help dispel the feeling that unlawful violence by law enforcement officers goes unpunished. The Assembly therefore calls on the authorities to grant access to this information”²².

Regional court judgments

A number of cases concerning deaths following an alleged unlawful use of force in law enforcement operations in France have come before the European Court of Human Rights. Excluding deaths in custody involving restraint techniques, the following cases are known:

Taïs v France (2006): The intoxicated victim was arrested and placed in a drunk tank at the Arcachon police station. He was found dead in the early hours of the morning. The ECtHR held that given discrepancies in the medical report and autopsy, the authorities had failed to provide a plausible explanation for the cause of the victim’s injuries that led to his death and which could only have been caused while he was in custody. In addition, the police officers showed inertia in the face of the victim’s physical and moral distress. Both these failings breached Article 2.

Saoud v France (2007): The police received a telephone call asking them to intervene at the victim’s family home, where he was attacking his mother and two sisters. The victim refused to open the door to the flat when requested by the police and attacked one of his sisters. The police entered by the balcony and fired two kinetic impact projectiles [the report does not specify which sort of projectile]. The victim managed to seize one officer’s hand gun and fired four shots at floor level before he was disarmed. He was then handcuffed with his arms in front of his body, while the body weight of the officers was used to keep him pinned to the ground on his stomach. Held for thirty minutes in this position, he died of cardiorespiratory arrest due to “slow asphyxiation”. The ECtHR decided that the use of force at the initial moment of intervention and capture did not breach Article 2(2), but that the subsequent restraint technique did breach Article 2.

Guerdner and others v France (2014): While under arrest and being questioned in a gendarmerie headquarters, the victim jumped out of a window to try to escape. A gendarme fatally shot the victim as he ran away in darkness. The ECtHR held that the regulations covering the use of force were unclear but were compatible with Article 2 ECHR. It held that the use of lethal force was not absolutely necessary in the circumstances and so breached Article 2(2). Although the investigation into the use of lethal force by a gendarme was carried out by the gendarmerie, it was found not to breach the Article 2 requirements for an investigation.

Toubache v France (2018): The victim was a passenger in a car, the driver of which did not stop when ordered to do so by a gendarme. As the car was being driven away, the gendarme shot at the vehicle to try to stop it, but in so doing killed the victim. The ECtHR held that the use of force was not absolutely necessary in the circumstances and so breached Article 2(2).

Mendy v France (2018) (admissibility decision): The victim was suffering from a psychiatric illness and was chasing another person with a knife. He was shot dead by a police officer who formed the view that the victim represented an immediate danger to the life of another person. The ECtHR held that the use of lethal force in the circumstances was clearly within the scope of the requirements for absolute necessity under Article 2(2), so there was no case to answer.

Semache v France (2018): The victim died after being arrested by the police when the car in which he was travelling was stopped for erratic driving. The victim was forcefully restrained using a chest compression technique. The ECtHR held that although the force used was proportionate in the circumstances, the victim’s situation was handled negligently by the authorities, who failed to do what could reasonably be expected of them to prevent the risk of death to which he was exposed, and so breached Article 2.

Boukrourou and others v France (2018): The victim suffered from a psychiatric disorder and became annoyed and agitated in a pharmacy when he could not obtain his medication. The police intervened. Faced with his refusal to comply with their instructions and his physical resistance, the police officers punched him to subdue him. The victim died shortly afterwards. The ECtHR decided that a causal connection between the police use of force and the death could not be established, so there was no breach of Article 2. As medical assistance was called rapidly, that also meant Article 2 was not breached. However, the ECtHR held that the violent, repeated and ineffective actions of the police officers constituted an affront to human dignity and were thus contrary to Article 3.

Bouras v France (2022): During transfer, the victim tried to take a gendarme’s service weapon and escape. The driver then stopped and tried to subdue the detainee, before finally opening fire and killing him. The ECtHR held that this was a situation of self-defense and so within the scope of Article 2.

National court judgments

Yassin Aibeche (2016): A fight had broken out in 2013 in a Marseille grocery store between the police officer, who was intoxicated outside working hours, and Yassin Aibeche, a high school student. The policer officer used his service firearm and killed the victim. In 2016, he was sentenced by the Bouches-du-Rhône Assize Court to twelve years’ imprisonment.

Amine Bentounsi (2017): The victim, an unarmed fugitive trying to escape from the police, was shot in the back in Noisy-le-Sec in 2012. The police officer who fired the shot pleaded self-defence, but received a 5-year suspended prison sentence.

Mustapha Ziani (2017): During an intervention in 2010, a Marseille police officer shot a man in the chest with a KIP launcher. He received a 6-month suspended prison sentence in 2017.

Dieng v France (2020): The victim died in a police bus after being immobilized and pressed to the ground while resisting arrest. In 2017, the victim’s family lodged an application with the ECtHR. In 2020, the French State reached an out-of-court settlement by virtue of which the claimants agreed to renounce all further claims against France in relation to the facts at the origin of this claim, while the government undertook to pay them the sum of 145,000 euros.

Najim A. (2021): In 2018, a driver fleeing a traffic stop at the Gallargues toll booth was shot dead by a gendarme. In 2021, the gendarme was found guilty by the Judicial Court of Nimes for involuntary manslaughter and sentenced to a 1-year suspended prison sentence and a 5-year ban on carrying a weapon.

Yannick Locatteli (2021): Responding to a suspected burglary in 2018 in Guadeloupe, a gendarme fired seven bullets into the victim’s vehicle as it was reversing, fatally wounding him. In 2021, the gendarme was sentenced by the Judicial Court of Basse-Terre to 5 years’ imprisonment, 3 of which were suspended, for involuntary manslaughter.

Amadou Koumé (2022): In March 2015, the police were called by a waiter in a Paris bar, who reported an agitated man. Three police officers intervened to arrest the victim. Investigations established, largely thanks to CCTV footage, that the victim had been strangled, first standing up for around ten seconds, then lying on the ground for a longer period. The victim was also held down by a police officer, who applied pressure with his knee to the victim’s lumbar vertebrae. Once handcuffed and restrained by a plastic collar placed around his legs, the victim was left to lie face down on the floor on his stomach. In 2022, the three police officers received a 15-month suspended prison sentence from the Judicial Court of Paris for involuntary manslaughter.

Cyril Cozic (2022): In January 2014, the victim, who was intoxicated with alcohol, was taken into custody following a hit-and-run road accident. The victim was placed in a drunk tank and was found dead the next day. During the night, the temperature had dropped to 8°C. In 2022, the Judicial Court of Versailles gave the gendarme a six-month suspended prison sentence for involuntary manslaughter.

Oversight bodies

The National Police are overseen by the IGPN. The IGPN is the result of the 1986 merger of the previous Inspection Générale de la Police Nationale (IGPN), created in 1884, and the General Services Inspectorate (Inspection Générale des Services or IGS), created in 1854. The IGS’s jurisdiction was limited to the Paris police prefecture (i.e. Paris intra muros and the departements (administrative areas) of the Petite Couronne, including the Paris airports of Orly and Roissy).

This body carries out general inspections and audits of the police and investigates the conduct of police personnel — its activities are outlined on the Interior Ministry website. It publishes annual reports on its activities, including data on the use of force.

From 2013 onwards, the annual reports were expanded and presented at a press conference. In 2017, the IGPN announced the creation of official statistical figures in order to track deaths and injuries. The 2020 report introduced more information on fatalities, including a brief description of the incident, date and location.

The National Gendarmerie are overseen by the IGGN. Previously, the Inspectorate of the National Gendarmerie (Inspection de la Gendarmerie nationale or IGN), created in 2002, was under the supervision of the Defence Ministry. In 2011, the IGN was replaced by the IGGN when the Gendarmerie was attached to the Interior Ministry.

The IGGN carries out general inspections and audits of the gendarmerie and investigates the conduct of gendarmerie personnel — its activities are set out on the Interior Ministry website. It publishes annual reports on its activities since 2019, including data on the use of force since 2020.

Part 2: Policies and Procedures

Data collection and publication by official agencies

1. Are the number of deaths following any police use of force
	Collected?	Good, Robust
	Accessible through existing publicly available information?	Good, Robust
	Is this a legal requirement?	No Provisions
	Can such information be requested from the authorities when not publicly available?	Not applicable
	If one can request it, what is the likelihood this information would be released?	Not applicable
	Reason(s) why unknown or not applicable	Already in the public domain
2. If published, to what extent can the number of deaths be readily determined from official statistics?		Good, Robust
3. Is it possible to identify specific individuals killed in official records?		Partial, Medium
4. Is demographic and other information for the deceased
	Collected?	No Provisions
	Accessible through existing publicly available information?	Limited, Poor
	Is this a legal requirement?	No Provisions
	Can such information be requested from the authorities when not publicly available?	No Provisions
	If one can request it, what is the likelihood this information would be released?	No Provisions
5. Is demographic and other information on officers in use of force incidents
	Collected?	No Provisions
	Accessible through existing publicly available information?	Limited, Poor
	Is this a legal requirement?	No Provisions
	Can such information be requested from the authorities when not publicly available?	No Provisions
	If one can request it, what is the likelihood this information would be released?	No Provisions
6. Is information on the circumstances
	Collected?	Partial, Medium
	Publicly available?	Partial, Medium
	Is this a legal requirement?	No Provisions
	Can such information be requested from the authorities when not publicly available?	Good, Robust
	If one can request it, what is the likelihood this information would be released?	Unknown
7. Is information about the type(s) of force used
	Collected?	Good, Robust
	Accessible through existing publicly available information?	Good, Robust
	Is this a legal requirement?	No Provisions
	Can such information be requested from the authorities when not publicly available?	No Provisions
	If one can request it, what is the likelihood this information would be released?	No Provisions

Data quality of official sources

8. How reliable are the sources used to produce official statistics about deaths?		Partial, Medium
9. Are there mechanisms for internal quality assurance / verification conducted		Unknown
10. Is the methodology for data collection publicised?		Partial, Medium
11. How reliable are the overall figures produced?		Partial, Medium

Data analysis and lessons learnt

12. Do state or police agencies analyse data on the use of lethal force, to prevent future deaths?		No Provisions
13. Is there evidence that state/ police agencies act on the results of their analysis, including applying lessons learnt?		No Provisions
14. Are external bodies are able to reuse data for their own analyses?		Limited, Poor
15. Do external, non-governmental agencies collect and publish their own statistics on deaths following police use of force?		Good, Robust

Investigations by official agencies

16. Is there a legal requirement for deaths to be independently investigated?		Good, Robust
17. If so, which organisation(s) conduct these investigations?		IGPN resp. IGGN (depending on whether a Police resp. Gendarmerie officer was involved), and the Public Prosecutor
18. In the year in question, how many deaths following police use of force have been investigated by the organisation(s) specified in question 17?		Probably 100%
19. Are close relatives of the victims involved in the investigations?		Partial, Medium
20. Investigation reports into deaths
	Are publicly available?	No Provisions
	Give reasons for the conclusions they have reached?	No Provisions
	Is this a legal requirement?	No Provisions
	Can such information be requested from the authorities when not publicly available?	Good, Robust
	If one can request it, what is the likelihood this information would be released?	Partial, Medium
21. Is there information available on legal proceedings against agents / officials, pursuant to deaths?		Partial, Medium
23. Is there information available on disciplinary proceedings against agents/ officials, pursuant to deaths?		Partial, Medium
24. Number of prosecutions against agents / officials involved in the last ten years?		Exact number unavailable
25. Number of convictions against agents / officials involved in the last ten years?		At least 8 (5x Police, 3x Gendarmes)
26. Number of prosecutions against agencies involved in the last ten years?		No legal proceedings against Police nor Gendarmerie.
27. Number of convictions against agencies involved in the last ten years?		No legal proceedings against Police nor Gendarmerie.
28. Number of cases in which states have been found to have breached human rights law on the use of lethal force?		4 ECtHR cases: Saoud v France (2007) and Semache v France (2018), both Police; Guerdner v France (2014) and Toubache v France (2018), both Gendarmerie.

Detailed elaboration

Data collection and publication by official agencies

Recording incidents of persons killed in connection with law enforcement activities in France is not required by law. The IGPN has published annual reports since 2013 but has been collecting and publishing data about deaths and injuries connected with police operations since 2017. The IGGN has been collecting data about deaths and injuries connected with gendarmerie operations since 2019 and publishing reports since 2020. The data is made publicly accessible in the IGPN and IGGN annual reports, which are available online.

The IGPN has subsequently published its reports for 2020, 2021 and 2022, and the IGGN has published its first reports covering the years 2019, 2020, 2021 and 2022. These IGPN and IGGN reports demonstrate incremental increases in the amounts and types of data made available about National Police and National Gendarmerie uses of force, and injuries and deaths that have occurred in relation to them. The availability of new types of data, and especially the first published reports from the IGGN, are significant developments. However, concerns remain about the clarity and comparability of this data, and the extent to which patterns, trends and key information are obscured in an abundance of details and different ways of presenting them.

IGPN data collection and publication

The IGPN Annual Reports are currently available online: https://www.interieur.gouv.fr/Publications/Rapports-de-l-IGPN. For 2018 and 2019, the IGPN recorded 15 deaths and 106 injuries in 2018, and 19 deaths and 117 injuries occurring during or after police operations in 2019. In 2018 the Annual Report only provided total numbers of deaths and injuries, whereas in 2019 the Annual Report provided a breakdown of ‘causes of death’ into six categories by apparent cause of death: electric discharge pistol (TASER), individual sidearm, other firearm (including rifle or pistol), suicide or self-harm, ‘other’ (including accident, fall or escape), and incidents not involving a weapon. In those reports the IGPN did not publish details about the circumstances or geographical location of the deaths and injuries, nor did it publish names or biographical details about people killed or injured.

There is a notable increase in the amount of contextual explanation from the 2020 report onward. The IGPN has formalized data presentation and reproduced the format below in reports from 2020 to 2022.

• A bar chart indicating the context in which deaths occurred. This is broken down into six categories: non-terrorism-related judicial operations, terrorism-related judicial operations, police interventions, transfer and escort, other incidents away from police stations, incidents inside police stations. Some of the terminology used to identify these contexts is vague, however, and it is not possible to discern what exactly is meant by labels such as ‘other incidents’ and ‘other interventions’. By way of comparison, this information was not included in previous reports.
• A ring or pie chart indicating the ‘origin’ or cause of death. This is preceded by the caveat that “the record of deaths includes persons who were in contact with the police or in their custody. This is why, for example, these data include deaths that occurred inside police stations and that were the result of a poor state of health”. The chart includes five categories: use of a weapon, deaths linked with the person’s state of health, deaths linked with the person’s behaviour (suicide, drowning etc), use of physical force, traffic accident.
• A bar chart or textual summary indicating the context in which injuries occurred. This is broken down into eight categories: maintaining public order, urban violence, non-terrorism-related judicial operations (e.g. investigations or searches), other police interventions, identity checks, transfer and escort, other incidents outside (i.e. away from) police premises, incidents inside police premises. Sometimes this data is provided without visual representation: for example, the 2019 report provides some of this information as a mixture of numbers of instances and percentages of the total for the year.²³
• A comparative bar chart showing the numbers of injuries by cause in 2019 and 2020 (shown as 2019 figure/ 2020 figure). Notably, the heading for the chart flags that 58% of injuries were not caused by a weapon. The chart includes eight categories: other (accident, fall, escape) (12/9), no weapon used (45/45), use of defensive baton (12/9), use of KIP launcher (32/7), use of sting ball grenade (3/2), use of side arm (9/3), use of Taser (3/2), administrative vehicle (1/1). By way of comparison, the 2019 report provides this information in a different format.²⁴
• Lists of descriptive vignettes giving some details of all of the above deaths by category, specifying the date, location (town and department, i.e. administrative region), a brief summary of the circumstances, and the sex and age of the deceased (although the latter information was for some reason omitted from the 2022 report). This is a noteworthy innovation in the 2020 report in light of the concerns raised in Police Lethal Force and Accountability.

IGGN data collection and publication

The IGGN published its first annual report in 2019 and the reports for 2019, 2020, 2021 and 2022 are currently available online: https://mobile.interieur.gouv.fr/Publications/Rapports-de-l-IGGN. Notably, each of these reports publishes selected data not only from the year in question, but also previous years, going back as far as 2012 in some instances.

The IGGN Annual Reports from 2019 to 2022 become increasingly larger over time (from 39 pages in the 2019 report to 144 pages in the 2022 report)

The 2019 report summarises (p.15) three of the main reasons for internal investigations of officers’ conduct by the IGGN Judicial Investigation Office (bureau d’enquête judiciaire or BEJ) in 2019. These include illegal possession of ammunition or a weapon (3) and intentional violence with or without a weapon (31). The report then briefly outlines three types of incidents investigated by the BEJ:²⁵

• 2 incidents of a use of a weapon leading to death and 1 leading to serious physical injury (compared to 9 in total in 2018).
• 31 incidents of violence with or without a weapon by a person holding public authority (member of the National Gendarmerie), involving 12 incidents of stopping and questioning individuals in the street or in their home, 3 such incidents inside a Gendarmerie station, 12 during a public order operation and 4 by off-duty personnel.
• 12 incidents of violence by a person holding public authority for the maintenance of public order (member of the National Gendarmerie) leading to the injured person being unable to work for 8 days or more, or even permanently. Of these, 11 incidents were linked with the Yellow Vest demonstrations and the presumed use of force by a gendarme, including KIP launchers and grenades. The report flags that 50 such incidents were reported in 2018-19 on the IGGN online complaint platform but the BEJ was instructed to investigate in only 12. The report flags that the investigations that have been completed so far have found these uses of force to conform with the law.

No specific details are given about these various incidents or their context until the next report for 2020. There is a section entitled “Focus: uses of lethal weapons from 2017 to 2020” (pp.33–35). This section provides data about 17 incidents of lethal force involving gendarmes and investigated by the BEJ in this period, all of which related to the use of firearms (the report specifically states that other weapons including KIP launchers, grenades and Tasers did not cause deaths in this period). Of these the report specifies that 15 incidents involved a deliberate shooting, 1 related to an anti-terrorist operation, and that in ‘over half’ of the incidents the person killed had psychiatric or psychological problems, predominantly of a schizophrenic sort. The context and specifics of these incidents is summarised in detail in the text,²⁶ including type of operation, time of day, location, mental condition of the person killed, and details of the judicial investigation and outcome. The legal framework for the use of lethal force is summarised with text and graphics at the end of the section²⁷. At the end of the report are tables of data about numbers, circumstances and causes of deaths and injuries connected with National Gendarmerie activities and which led to a judicial investigation for 2018, 2019 and 2020.²⁸

In the report for 2021, the format for presenting figures seems to have stabilized. Since then, a description of the context of each death has also been included. In the report for 2022, there were 12 deaths (by context: 4 following the use of weapons, 8 during an operation or in custody; by cause: 3 due to the use of a firearm, 1 due to Taser, 2 to suicide or self-mutilation, 2 due to a National Gendarmerie vehicle, 4 due to ‘another reason’ such as an accident, a fall, sudden death, illness or escape) and 26 injuries listed (context: 24 during an operation; cause: 5 due to the use of armed force of which 2 due to a firearm, 2 due to a Taser, 1 due to truncheon, 8 due to a National Gendarmerie vehicle, 13 due to ‘another reason’ such as an accident, a fall, sudden death, illness or escape).

Further details about the 4 deaths involving a use of weapons are provided in a table (p.92) specifying the date of death, location, type of violence used by the deceased person and a brief outline of the circumstances; details about the other 8 deaths are provided in a table (p.93) specifying date, location and circumstances. Further details about persons injured follow (pp.94–97), with tables for injuries following the use of a firearm, injuries during an operation not involving a firearm and injuries suffered in custody, injuries following use of a ‘weapon of intermediate force’ (two Tasers and a truncheon), and injuries following unarmed action by gendarmes.

Data quality of official sources

In addition to the issues of data clarity and comparability detailed in the previous section, there are some data quality concerns pertaining to deaths that are recorded in broad categories.

The IGPN and IGGN do not provide clear information on their methods of gathering, analysing, checking and presenting data. There also seem to be some inconsistencies when it comes to chase fatalities. Their selection —or not— is not always justified. As Basta points out²⁹, the fatal accident in Montchanin (Saône-et-Loire) that claimed the life of a 17-year-old youth for refusing to comply with a gendarme’s instructions is not referenced by the IGGN, while a death in Jardres (Vienne) in July 2021, under similar circumstances, is counted. The IGPN also seems to take this type of death into account (as in the 2022 report, for example). However, Basta also points to cases not selected by the IGPN, such as that of a man who drowned in the Louet river, near Angers, while fleeing from the police, in 2021.

Furthermore, descriptions are not always as factual as they claim to be, and may omit certain details, often when this is detrimental to the police or the gendarmerie. For example, when describing the death of Cédric Chouviat in 2020, the use of the chokehold and the belly tackle were omitted, even though it was one of the key elements of the public controversy following the death³⁰.

IGPN data methodology

Concerns about methodology in the IGPN reports include the way in which the tables providing data on the context and cause of deaths refer to broad categories such as ‘non-terrorism related judicial operations’, ‘interventions’ and ‘other incidents’.

The IGPN states that its data is produced from statements and declarations from police services, but no information is provided about who submits and compiles the data, how they are trained, what their methods are or whether and if so how the data is checked.

When the IGPN started to list the deaths and injuries of civilians, the initiative was not widely known by the police services. As a result, there was not much feedback from the various police units that may have been involved in the deaths of civilians. The first reports 2018–2019 were therefore incomplete. The IGPN had to do a lot of internal communication with police units to improve feedback³¹.

Overall, the IGPN reports for 2020-2022 include some more detail and contextual information than the two previous reports for 2018 and 2019. They fill some of the gaps flagged in the Police Lethal Force and Accountability 2018 country report on France and the National Police, but some of the data is nevertheless hard to interpret, break down, cross-reference and compare across reports for previous years. That being said, working solely on the basis of declarations seems to preclude any claim to exhaustiveness, as the previously cited gaps tend to show.

IGGN data methodology

The IGGN gives no details of its methodology in its reports.

Data analysis and lessons learned

The IGPN and IGGN annual reports contain no recommendations for preventing future deaths or questioning police practices and techniques.

Nor do the two institutions seem interested in describing the victims and, consequently, proposing a reflection on LEO actions based on the findings. The IGGN reports offer no information about victims. It is possible to guess the gender of the victims from the IGPN reports, but the latter do not offer any aggregated data. Age was also indicated in the descriptions of the 2020 and 2021 reports, but has unfortunately disappeared for the 2022 report.

French authorities are prohibited from collecting statistics on race or religion. This is a political choice, intended to give the impression that all French citizens are equal before the French authorities. However, in practice it usually acts to prevent the study of possible racial discriminations. That said, it is possible —and could be insightful— to have demographic data on the nationality of the civilians killed and that of their parents.

Investigations by official agencies

In France, judicial investigations are divided into several stages.

Once the death has been established, the public prosecutor’s office must open a preliminary investigation, conducted by the IGGN or the IGPN. At the end of the preliminary investigation, the public prosecutor may decide either to close the case with no further investigation (classement sans suite) or to open a judicial investigation (ouverture d’une information judiciaire).

In the latter case, an examining magistrate (juge d’instruction) is appointed to continue the investigation, both incriminating and exonerating, in order to determine the existence or absence of an offence. At the end of the investigation, the judge may either dismiss the case if he or she considers that there is insufficient evidence to initiate criminal proceedings (ordonnance de non-lieu), or order that the case be referred to the trial court if he or she considers that sufficient charges have been brought.

In the latter case, one or more trials may be held, at the end of which the law enforcement officer is either convicted or acquitted.

At any time, the victim’s family can lodge a complaint and/or register as a civil party for access to the investigation file. They can also contest if the prosecutor did not open a judicial investigation or if there was a dismissal of the case.

In addition to the judicial aspect, the IGPN and IGGN can also conduct administrative investigations and apply disciplinary sanctions.

The Defender of Rights (Défenseur des droits) can sometimes decide to look into a case, carrying out an investigation (with the possibility of interviewing people involved and having access to judicial documents). He then issues a “decision”, in which he recalls the facts, draws conclusions and proposes recommendations (whether disciplinary sanctions or more general recommendations relating to the application of the law and police action). That said, those recommendations are rarely followed up. In the 2019 annual report, the Human Rights Defender points out that he has requested disciplinary proceedings in 36 cases since 2014. However, none of his requests have been considered³².

The involvement of close relatives in investigations depends on the investigating judge in charge, whether they deem it necessary: There are no commonly shared practices in this area in France.

Legal proceedings, prosecutions and convictions

National Police

The Annexes to the IGPN 2020 report include data about investigations into alleged police misconduct:

• Annexe 2 (p.82) lists the number of alleged criminal offences for which a legal investigation was started. This includes ‘offences against the person using force’, of which there were 532 in 2020. This is not further broken down in terms of injury, death, type of force used or final decision and is not cross-referenced with the data about deaths and injuries in chapters 2 and 3.
• Annexe 6 (pp.82–85) lists instances of ‘professional fault’ by type. This includes a category (p.84) of ‘use of force or constraint’, of which there were 42 instances. These consisted of disproportionate use of force or constraint (21), disproportionate use of force or constraint without a weapon followed by injuries (2), disproportionate use of force or constraint involving a weapon of intermediate force or other means of intermediate force followed by death (1), disproportionate use of force or constraint involving a weapon of intermediate force or other means of intermediate force followed by injuries (18). However, no contextual details are given and these figures are not cross-referenced with the data on deaths and injuries presented in chapters 2 and 3.

The Annexes to the IGPN 2021 report similarly include data about investigations into alleged police misconduct:

• Annexe 3 (p.112) lists the number and type of alleged criminal offences for which a legal investigation was started. This includes ‘offences against the person using force’, of which there were 510 in 2021. As in the 2020 report, this is not further broken down in terms of injury, death, type of force used or final decision and is not cross-referenced with the data on deaths and injuries in chapters 2 and 3.
• Annexe 6 (pp.113-116) lists instances of ‘professional fault’ by type. This includes a category (p.115) of ‘use of force or constraint’, of which there were 42 instances (the same number as in 2020). These consisted of disproportionate use of force or constraint (20), disproportionate use of force or constraint without a weapon followed by death (2) or injuries (8), disproportionate use of force or constraint involving a weapon of intermediate force or other means of intermediate force followed by death (0) or injuries (9); disproportionate use of force or constraint involving a firearm followed by death (0) or injuries (3). However, no contextual details are given, and these figures are not cross-referenced with the data on deaths and injuries presented in chapters 2 and 3.

National Gendarmerie

Most of the information in the report for 2019 relating to the use of force, injuries and deaths consisted of broad outlines of BEJ investigations. The subsequent reports for 2020 and 2021 also included more detailed summaries of BEJ investigations of these matters. Consequently, while the publication of this data is to be welcomed, it is apparent that most of the data presented in IGGN reports relates only to matters referred for judicial investigation by the BEJ. As the IGGN report for 2020 notes at p.31 though, the BEJ is not the sole investigative body within the IGGN and investigations are allocated ‘after dialogue’ between the divisions that might be concerned. How exactly this allocation is determined, the implications thereof and numbers of investigations dealt with otherwise than by the BEJ are not specified, although for investigations into fatalities, the BEJ conducts the preliminary investigation.

According to the IGGN report for 2022 (p.9, 10 and 52), 783 internal judicial investigations were opened in the Gendarmerie in 2022, including 54 by the BEJ (4 due to use of lethal weapons). In 2022, 38 people were killed or injured as a result of gendarmerie action. These 38 incidents all gave rise to legal proceedings. 13 were dismissed and 24 investigations are still underway (one investigation concerns both a deceased and an injured individual) (p.89).

The overall volume of disciplinary cases in the Gendarmerie Nationale is on the rise. It represents 2,992 sanctions in 2022, compared with 2,809 in 2021 (i.e. +183 sanctions). A detailed and quantified list is provided in the appendix of the report (p. 140–141).

Non-official sources

A number of non-official sources also provide statistics on deaths following police contact, or in police custody. These may include deaths following police use of force, though such deaths are not always readily identifiable. Sources include, but are not limited to, the following:

• An important non-official source of data about deaths connected with law enforcement agency activity in France is the work of the Basta newspaper. This independent media publication established in 2014 a database of cases of ‘deaths by the police’. This database is published on the newspaper’s website and includes short analytical texts together with various visual representations of the available data.
• Basta also documented 213 cases of lethal police intervention with known judicial outcomes, from January 1977 to June 2020. They show that two-thirds of cases never go to trial. Of the 213 cases for which the outcome is known: 39% were closed (no judicial investigation opened), 28% were dismissed (the judicial investigation did not lead to a trial) and 33% went to trial. For cases that have gone to trial, 18% of LEOs received suspended prison sentences, 10% were discharged or acquitted and 5% were sentenced to prison.
• In his thesis work and using available media data³³, Paul Le Derff attempted to describe fatal police interventions and their victims between 1990 and 2016 in France (excluding fatal police interventions linked to terrorism). A typology of eight categories was created to list the 360 fatal police interventions: use of a firearm against an unarmed person (n=34); arrested or taken into police custody (n=33); use of force against an armed person (n=84); fatal shootings at a moving vehicle (n=47); fatal escapes (n=80); deaths and suicides at the police station (n=37); fatal use of firearms at home (21), and others (n=24). The study also shows that young men from low-income neighbourhoods and with contemporary links to immigration are over-represented among the 393 victims.

Part 3: Comparative indicators

I-1a. CK: Number of civilians killed by law enforcement agents on duty, by gunshot	10
I-1b. CKt: Number of civilians killed by law enforcement agents, regardless of means and whether or not on duty	26
I-1c. CW: Number of civilians wounded by law enforcement agents on duty, by gunshot	12
I-1d. CWt: Number of civilians wounded by law enforcement agents, whether or not on duty and regardless of means	133
I-2. CK per 100000 inhabitants	0.01
I-3. CK per 1000 law enforcement agents	0.04
I-4. CK per 1000 arrests	Data unavailable
I-5. CK per 1000 weapons seized	Data unavailable
I-6. AK: Number of law enforcement officers unlawfully killed on duty by firearm	0
I-6b. AKt: Number of law enforcement officers unlawfully killed, whether or not on duty and regardless of means	17
I-7. AK per 1000 agents	0
A1. Percentage of homicides due to state intervention	1%
A2. Ratio between CK and AK	10/0
A3. Civilian lethality index: Ratio between CK and CW	0.8
A4. Lethality ratio: Ratio between Civilian lethality index and law enforcement agents lethality index	0.8/0
A5. Average number of civilians killed by intentional gunshot, per incident	0.5

Number of civilians killed and injured

Civilian deaths

Analysis of official data indicates that in the 2019 reporting period at least 26 deaths occurred following the use of force by on-duty police and gendarmes. These were comprised of:

• 8 deaths classified as fatal shootings, comprising 6 recorded by the IGPN and 2 by the IGGN;
• Two fatal shootings linked to terrorism are not considered by the IGPN in 2019.
  • Mickaël Harpon was a police officer who committed a knife attack against his colleagues on October 3, 2019 inside the Paris police headquarters. The Islamic State later claimed responsibility for the attack.
  • On March 5, 2019, Michaël Chiolo, a prison inmate, and his girlfriend, Hanane Aboulhana, attacked and stabbed two guards. The couple retreated to a family living unit in the prison. The RAID police unit attacked, killing Hanane Aboulhana. The Minister of Justice refers to a ‘terrorist attack’ linked to the alleged radicalization of the prisoner.
• 18 deaths classified as occurring due to means other than gunshot, including 13 recorded by the IGPN and 5 recorded by the IGGN

Civilian injuries

Both the IGPN and IGGN reports detail the number of civilians injured. The data is based on declarations and cannot therefore claim to be exhaustive. In 2019, the IGPN and the IGGN reported 133 civilians wounded by public security agents:

• 12 injuries were gunshots, comprising 9 recorded by the IGPN and 3 by the IGGN
• 121 injuries classified as occurring due to means other than gunshot, including 108 recorded by the IGPN and 13 recorded by the IGGN
• The civilians wounded by off-duty public security agents are not listed.

Number of law enforcement officers killed and injured

Only limited official data is available for the number of law enforcement agents killed on duty. The IGGN publish some data on gendarmes killed or injured while on duty — the IGGN report 2020 on 2019 indicates at p.73 that a total of 6 gendarmes were killed on duty in 2019, including in road accidents.

The IGPN does not publish this data. There is, however, an unofficial memorial to fallen police officers that lists eleven deaths in 2019³⁴.

None of these deaths were caused by firearms.

Indicators of use and abuse

Without comprehensive data, some indicators cannot be calculated (CK per 1000 arrests; CK per 1000 firearms seized).

Other indicators can be calculated using the figure of 26 civilian deaths connected with LEO activity as a starting point. The rate per 100,000 inhabitants (indicator I-2), of 0.4, was calculated using data from the French National Institute of Statistics and Economic Studies (Institut national de la statistique et des études économiques or INSEE) which estimated the population of metropolitan France on 1^st January 2019 as 64,821,954³⁵.

The rate of civilian killings per 1,000 LE agents (I-3), of 0.04, was calculated using INSEE data on the total number of police and gendarmes in 2019 of 224,000³⁶.

Turning to the abuse indicators, abuse indicator A-1 — the percentage of homicides due to state intervention has been calculated using the 880 homicides in France in 2019³⁷.

Proportionality of civilian killings

As detailed previously, there is very little existing data on civilians killed by the police. The IGPN sometimes —but not systematically— mentions the victim’s gender, while the IGGN does not.

France has always been administratively colour blind, and it would be illegal to propose an ethno-racial frame of reference to describe victims. However, it could be worthwhile to measure possible inequalities in terms of migratory origin. In this respect, the two agencies could draw inspiration from the typologies used in the TeO survey (Trajectoire et Origines, survey on the diversity of populations in France carried out by the Institute for National Demographic Studies³⁸) to better describe the civilians killed during police intervention.

Estimates, based on incomplete data, have shown that men were 11 times more likely than women to die during a police intervention. Likewise, immigrants, descendants of immigrants and foreigners are six times more likely to die during a police intervention than people without contemporary ties to immigration.

Summary and recommendations

This case study has assessed France across a range of internationally comparable indicators around the use of lethal force in policing and law enforcement, including: legal frameworks (including national and international law and regulations); policies and procedures around data collection, publication, analysis, investigations and lessons learnt; and comparative indicators for deaths following police use of force. Certain features of the system in France constitute relatively good practice when looked at internationally.

However, France needs to stop overlooking international (from Council of Europe to UN Committees) and national (Defender of Rights) bodies’ recommendations. Also, in recent years, a great deal of progress has been made in presenting official figures for people killed by the police, but it seems to us that this is still too much a case of political communication (creating figures with the sole aim of showing that the authorities are transparent and respect democratic principles) rather than a genuine open data policy designed to improve practice.

On legal frameworks (Part 1):

• The 2017 law seems to have introduced a double confusion. A practical confusion about the rules governing the use of firearms by police officers, and a confusion of standards about the compatibility with the ECtHR principles. This law should be clarified to ensure it is clear and compatible with the ECHR.
• While there have been several rulings against France by the ECtHR for breaches of the procedural dimension of Article 2 ECHR, and the Human Rights Defender has noted recurrent failings, the Interior Ministry and the Justice Ministry need to do more to ensure the independence and transparency of investigations into injuries and deaths connected with National Police or National Gendarmerie activities.

On comparative indicators and data quality (Part 3) we recommend that both the IGPN and the IGGN:

• Provide more context about the fatalities, i.e. not to omit the different versions of the events, if there are several.
• Provide more systematic data on gender.
• Publish data about the nationality of victims and, if possible, the nationality of their parents, in a manner similar to what the TeO survey proposed.
• Clarify the methodology used in gathering data. This is especially important for the IGGN, which gives no indication on this subject.
• Improve the presentation of disaggregated data on deaths to avoid broad categories and facilitate identifying the context and cause of death. Although a degree of stability seems to have been achieved, enabling year-on-year comparisons to be made, there is still room for improvement, both in terms of the agencies’ individual data presentation and the need for a harmonised approach across both agencies. In the future, improved data presentation would make it easier to identify changes over the years, question practices and techniques, and facilitate comparisons between the two law enforcement agencies.

On analysis, investigations and the useability of data, we recommend that:

• Both the IGPN and IGGN do more to analyse data on deaths and injuries, and on the use of force (involving firearms, other weapons or due to other forms of interaction) so as to make evidence-based recommendations about National Police and National Gendarmerie operations with a specific view to reducing the incidence of harm and death.
• The IGGN in particular, and to a lesser extent the IGPN, provide clearer information about the outcomes of investigations than they did in the past. Both agencies now provide a little more data about internal investigations and disciplinary measures but this could still be improved, and it would be useful to have more specific data about the outcomes of the judicial investigations.
• The IGGN in particular finds an appropriate balance between detail and readability in its reports. We note that its reports are becoming increasingly dense and, although this may be seen as a good sign in terms of apparent openness, we also note that the distinction between communication and information seems to be increasingly blurred.

We recommend that the police, gendarmerie and/or other relevant bodies:

• Collect and publish statistics on deaths of law enforcement agents and staff.
• Collect and publish data on the types of force used in any such fatal incidents.

References, data sources and downloads

1 Étude d’impact du projet de loi relatif à la sécurité publique, available online at https://www.senat.fr/leg/etudes-impact/pjl16-263-ei/pjl16-263-ei.html#fnref11. The entire legislative career of this law is available online at https://www.assemblee-nationale.fr/14/dossiers/securite_publique.asp.