Japan Revises Its National Anti-Stalking Strategy

A warning is a piece of paper. Obsession moves.

That mismatch sits at the heart of Japan’s revised national anti-stalking strategy. For a quarter-century, the country has repeatedly widened its law after technology or tragedy revealed a gap: first repeated emails, then social-media messages, then covert GPS units, then tiny consumer tags and third parties who might disclose a hidden address. Each amendment defined more conduct and gave police more authority. Yet a prohibited person could still cross a city with a knife.

On July 28, the interministerial council on stalking countermeasures issued the fourth version of Japan’s Comprehensive Measures Against Stalking. The same day, the Cabinet’s ministerial crime council adopted emergency measures against fraud and stalking. The immediate backdrop was a fatal March attack in an Ikebukuro commercial complex: according to the government strategy, the man had already received criminal punishment for stalking-related offenses and had been subject to a prohibition order before killing the woman he had targeted.

The revision is an attempt to make the system continuous. Police are to use a nationwide common system for case management and information sharing. Every perpetrator under a prohibition order is to be contacted so officers can assess current circumstances and the intensity of fixation. Government agencies will study electronic monitoring that could warn a victim when a high-risk perpetrator approaches, as well as a system that could require some perpetrators to undergo treatment or counseling.

Those last two ideas made the headlines. They are also the easiest to misunderstand. Japan has not authorized police to fit stalkers with GPS bracelets. It has not made counseling compulsory. The document directs agencies to study legal, technical and operational design, foreign systems, human-rights safeguards and the capacity of treatment providers. What changed on July 28 was national policy direction—not the legal status of a tracking order.

Six pillars, not one tracking device

The nine-page strategy organizes the response into six areas. The first strengthens consultation and coordination. Police headquarters and local stations are to operate under a clear command structure, increase specialist staffing and training, and run a common nationwide system so a case does not fragment when a victim or perpetrator crosses a prefectural boundary. Women’s support centers, domestic-violence centers, schools, prosecutors, legal-aid offices, local victim services and mental-health centers are expected to connect rather than send a victim from door to door.

The second pillar protects information. Public records can betray a person who has escaped: a resident register, electoral roll, vehicle record, pension or insurance procedure, property document, school inquiry or mail-forwarding record can become a route back to an address. The strategy orders continued strict handling across ministries and local government. Police may warn a person or organization believed likely to give protected information to someone considered at risk of stalking, and ask that it not be supplied.

The third covers escape and rebuilding. It includes emergency protection at night and on holidays, police subsidies for temporary accommodation, support from women’s shelters, help finding long-term housing and work, access to public rental housing, and legal assistance. Safety is not achieved merely by spending one night elsewhere; a victim may have to change a home, job, school, telephone number and daily routine while absorbing costs created by somebody else’s conduct.

The remaining pillars cover research and education, perpetrator intervention, and funding. Schools are to teach healthy relationships, privacy and online risk. Police and health professionals are to develop better ways of responding to fixation. Prisons and juvenile institutions are to establish stalking-specific programs. The national and local governments are asked to train personnel, build support systems and assist private organizations.

The numbers behind the urgency

Police received 22,881 stalking-related consultations in 2025, 3,314 more than the previous year. The 16.9% rise brought the total close to the record levels of the late 2010s and well above the 19,000–20,000 range of the preceding several years. A consultation is not the same as a proven crime, and counting methods and reporting behavior affect the trend. But the caseload measures how often the police system must assess fear, fixation and the potential for escalation.

Women were the reported victims in 19,880 cases, or 86.9%. People in their twenties accounted for 35.6% of identified direct victims, followed by those in their thirties at 20.6%. Former or current dating partners formed the largest relationship category, 8,910 cases or 39%. Yet 2,021 cases involved a person with no prior acquaintance, and workplace connections, friends, spouses, customers and fans also appeared. Stalking is strongly gendered, but it is not confined to one relationship or age group.

Police recorded 12,181 cases involving following, ambush or similar conduct; 7,548 demands for meetings or a relationship; 4,911 silent or repeated calls, emails or messages; and 4,018 involving violent words or behavior. Categories overlap because one case can contain several behaviors. Covert location tracking was smaller in raw numbers but technically important: police counted 118 cases of non-consensual location acquisition using GPS-type equipment and 393 cases of attaching or placing such devices.

Administrative and criminal action also rose. Police issued 1,577 warnings and a record 3,037 prohibition orders, including 1,840 emergency orders. There were 1,546 arrests or cleared cases under the anti-stalking law and 2,171 related Penal Code or other statutory offenses, both post-enactment highs. The latter included two completed murders, 14 attempted murders, 153 assaults causing injury and 437 unlawful entries.

Numbers can show workload and harm; they cannot identify which case will turn lethal. That is the operational dilemma. Many incidents stop after advice, warning or arrest. A small share escalate rapidly. False reassurance is dangerous, but treating every case as equally imminent can exhaust the specialist attention needed for the most dangerous ones. The nationwide system will matter only if it improves structured risk assessment, review and action—not merely data entry.

Okegawa: the murder that changed the law

Modern Japanese anti-stalking law begins with a failure. In 1999, a 21-year-old university student was killed near Okegawa Station in Saitama Prefecture after months of harassment by her former boyfriend and people acting with him. She and her family had gone to police. The subsequent investigation exposed serious mishandling of their complaint and efforts within the police to disguise that failure.

The case transformed stalking from something often minimized as a private romantic dispute into a public-safety and institutional-accountability problem. In May 2000, lawmakers enacted the Act on Regulation of Stalking Conduct and Assistance to Victims; it took effect on November 24. The statute allowed warnings and prohibition orders and criminalized repeated conduct driven by romantic or affectionate feelings, or resentment when those feelings were not reciprocated.

That motive element remains important. Japan’s anti-stalking statute does not cover every pattern of harassment commonly described as stalking in ordinary speech. Conduct arising from business conflict, neighborhood hostility or other motives may instead be addressed through assault, threat, trespass, obstruction, nuisance ordinances or other laws. Legal classification determines which preventive powers are available.

The original law was a breakthrough, but it was written for an era of landlines, letters and physically waiting outside a home. The next quarter-century became a race between statutory verbs and changing methods of intrusion.

A law rewritten by cases and technology

The 2013 amendment added repeated email transmissions and made it easier for authorities outside the victim’s home prefecture to act. The debate followed cases in which threatening electronic pursuit sat outside the wording of a law built around telephone calls and physical presence. But communications changed again: a platform message could terrorize as effectively as an email.

The 2016 reform added repeated messages through social networks and similar services, as well as loitering near a home, workplace or school. It removed the requirement that a victim formally complain before stalking could be prosecuted, raised penalties, expanded information and victim-assistance duties, and allowed prohibition orders without always waiting for a prior warning. The reform followed the near-fatal stabbing of a young performer in Koganei after she had sought police help over online messages.

Location technology created another gap. A perpetrator could hide a GPS unit on a vehicle and retrieve its movements remotely. Courts found that the older statutory idea of “watching” near a residence or other place did not necessarily reach location acquisition performed elsewhere. The 2021 amendment directly regulated non-consensual GPS location acquisition and secretly attaching a device; the strategy was revised in 2022.

By then, consumer products had moved on. Small Bluetooth-based loss-prevention tags were inexpensive, long-lived and dependent on networks of nearby phones rather than a conventional GPS receiver. The 2025 amendment made their misuse explicitly regulable. It also created police warnings on their own authority, placed assistance expectations on employers and schools, expanded which public-safety commissions could issue orders, and empowered police from March 2026 to notify potential information providers that a recipient posed a stalking risk and request non-disclosure.

The two meanings of GPS

GPS appears on both sides of the 2026 strategy. The state is strengthening punishment for perpetrators who use location technology covertly against victims. At the same time, it is considering using location technology openly against certain perpetrators to protect victims. The purposes and legal authority are different, but the juxtaposition reveals the central policy issue: who may know whose location, under what order, for how long, and with what safeguards?

The proposed monitoring system would detect when an electronically monitored perpetrator came close to a victim and notify the victim or other designated party. The strategy does not define the qualifying risk threshold, decision-maker, device, exclusion-zone radius, monitoring period, response to tampering, retention of location history or procedure for challenge. Those are not implementation details at the margin; they determine whether the system is lawful, accurate and useful.

A proximity alert can buy time. It can also transfer a frightening operational burden to the victim: a phone alarm announces that danger is close, but who responds and how quickly? Dense railway stations, apartment blocks and workplaces can generate complicated geofences. Batteries fail; signals drift; devices can be removed. Monitoring must therefore sit inside a response protocol with trained staff, rapid police action and a victim-led safety plan. A bracelet cannot become a technological substitute for protection.

The government also acknowledges the perpetrator’s rights. Continuous location monitoring is a significant intrusion and would require a clear statutory basis, necessity and proportionality, limits on use and retention, security against leakage, oversight and a path to review. A narrowly designed court-supervised measure for demonstrably high-risk cases is legally different from open-ended administrative tracking.

From “stop” to “change”

Prohibition tells a perpetrator what not to do. Treatment asks what would reduce the fixation, grievance, entitlement, impulsivity or other pattern that makes the conduct recur. The revised strategy orders police to contact everyone subject to a prohibition order, encourage treatment or counseling for all such perpetrators and other relevant cases, obtain help from regional psychiatrists and certified psychologists, and study how to raise uptake.

The scale of the gap is visible in the statistics. In 2025, police recorded 3,037 prohibition orders. They made 1,698 follow-up contacts with perpetrators and connected 233 people to counseling or treatment. These figures are not a single matched cohort and should not be divided as if every order and treatment occurred in the same cases. They nonetheless show why the government is asking how to move beyond voluntary referral.

Compulsion raises hard questions. Stalking is behavior, not one diagnosis, and a person can be dangerous without a treatable mental illness. Programs must be matched to risk and need rather than assume that psychiatric care alone solves coercive control or calculated violence. Providers need training, capacity and safe ways to share risk information. Attendance is not the same as engagement; completion is not the same as reduced danger.

The strategy’s measured wording is therefore appropriate. It calls for research on mandatory participation, consideration of foreign systems and expansion of the medical and counseling “receiving side.” It also asks correctional institutions to build stalking-specific special guidance and treatment programs. A mandate without a credible program would produce compliance on paper, not safety.

The invisible perimeter around a victim

The deepest section of the strategy may be the least dramatic: information protection. A victim who relocates can remain exposed if separate institutions treat an address as ordinary administrative data. Japan’s plan names resident registration, family records, electoral rolls, vehicle registration, property and guardianship records, pensions, insurance, taxes, child benefits and mail forwarding among systems that require careful handling.

It also reaches private actors. Tracking-device manufacturers are to be pressed to design anti-abuse features. Private investigators are to verify the legitimacy of requests. Employers and schools now have a statutory duty to make reasonable efforts to assist victims, and they are warned not to reveal a workplace, schedule, classroom or child’s location.

This is an important change in the model of responsibility. A stalker may exploit whichever institution has the weakest procedure. Protection is only as strong as the receptionist, municipal clerk, school office, human-resources system, app alert and database query at the edge. The national strategy turns confidentiality from good manners into a safety control.

What success should look like

A policy this broad can appear successful because meetings were held, systems launched and leaflets distributed. The meaningful tests are harder. Do victims receive an informed response at the first approach, including when the conduct is not yet clearly criminal? Are high-risk cases reviewed across prefectures without delay? How many prohibited perpetrators are contacted, assessed and re-contacted? Do counseling programs reduce repeat conduct? Are address disclosures and tracking-tag misuse prevented? How quickly does an electronic-monitoring alert produce protection?

Government should publish measures with enough context to avoid perverse incentives. A rise in consultations can signal more offending, greater trust in police or both. More prohibition orders can mean more danger or better intervention. A low number of recorded information leaks may indicate success—or poor detection. Outcome evaluation needs repeat victimization, escalation, time to protective action, program completion, victim experience and independent review.

A strategy written between tragedies

Japan’s anti-stalking framework has rarely evolved in calm conditions. Okegawa produced the original law. Later killings and attacks exposed email, social-media, police-response and order-making gaps. Courts and consumer technology exposed the limits of location provisions. The March 2026 killing showed the most painful limitation of all: a person can be known to the system, punished and prohibited, yet remain dangerous.

The July revision does not promise that every act can be predicted. It does something more credible. It recognizes that protection requires a chain: an early listener, an accurate risk record, a secure address, a safe room, a workable order, active follow-up, a response to fixation and accountability when any link fails.

Its ambition will be tested in ordinary moments, not ministerial meetings—when a student shows a teacher a stream of messages; when a municipal clerk receives an address request; when an employer sees someone waiting outside; when an officer inherits a file from another prefecture; when a prohibited person misses a counseling appointment; when a phone signals that a monitored person has crossed a boundary.

For 25 years, Japan has become better at naming the methods of stalking. The 2026 strategy asks a harder question: can the state recognize the movement toward violence before it reaches its destination?