Stalkerware and spyware triage for civil society
Stalkerware versus mercenary spyware and account compromise, Android features stalkerware abuses, detection tools with stated limits, and where to get help.
Updated
In short
Before any removal, the Coalition Against Stalkerware warns that removing stalkerware or making significant changes may be detected by the person monitoring and could increase abuse. Stalkerware is also only one route: in the one clinical study reviewed here (44 clients in one city, 2019; most no longer lived with the abuser and had reset devices, and the authors say the sample cannot estimate how common spyware is), potential spyware was found for 3 clients, against 14 with potential password compromise. Detection tools have stated limits: in a 2025 test on one Android phone, antivirus detection of 17 stalkerware apps ranged from 53% to 100%, and the report says in-app and push alerts risk exposing a detection to an abuser who monitors the device in real time; the MVT README says public indicators cannot show a device is clean, the SpyGuard README says it might miss malware that does not communicate during capture, and no independent detection-rate test of SpyGuard or iOS scanners was found. Help comes from specialist survivor-support and technology-safety organisations; the Access Now helpline serves civil society subject to its own eligibility check.
Safety before technique
The Coalition Against Stalkerware's survivor guidance says: "Removing stalkerware or other monitoring detection and/or making significant changes may be detected by the abuser and could increase the abuse and harassment. Only attempt to remove stalkerware if you believe it is safe to do so." It recommends a safety plan developed on a safer device with a survivor-assistance programme, and notes that deleting stalkerware also deletes evidence. It adds that anyone monitoring a device "would be able to see any searches for help or resources", and advises contacting help from a device the person has not had physical access to.
The same concern shapes the Coalition's guidance to security vendors: because removal can be detected, the device owner should be given a clear choice whether to remove a detected app. Apple's Safety Check documentation carries a similar warning: "Before making changes or deleting information, consider potential impacts to your safety and privacy."
These sources suggest that safety planning with a survivor-assistance programme comes before any removal or significant change, and that the choice rests with the device owner. This guide gives no legal advice; the legality of stalkerware varies by jurisdiction, as the AV-Comparatives report notes.
Three different problems
Stalkerware. The Coalition's definition is software, made available directly to individuals, that lets a remote user monitor another user's device without consent and without explicit, persistent notification, in a manner that may facilitate intimate partner surveillance, harassment, abuse, stalking or violence. Its survivor page says it requires no particular technical skill and is mostly installed on phones. Citizen Lab's 2019 report adds that surveillance software sold for ostensibly legitimate purposes can be repurposed, and that "even application functions which are included in mobile operating systems, such as those which help to find one's friends and colleagues, can constitute stalkerware under certain circumstances".
Government and criminal spyware. The Coalition's survivor page uses "spyware" as the umbrella term but says this class, tools used by governments and criminals, "is not available to the average person and is not covered by this website". For forensic checking of a phone against government-grade spyware, see Checking a phone for mercenary spyware.
Account and cloud compromise. Citizen Lab's 2019 report said that iOS stalkerware routinely depends on obtaining the target's iCloud password, whereas Android stalkerware involves installing an app; its full report notes exceptions among the iOS-targeting apps in its sample (it names FlexiSPY and mSpy in one passage). The Coalition's survivor page says some iPhone stalkerware reads iCloud backups, and that an iCloud Backup setting the user did not enable may indicate this.
The clinic study by Havron et al. (USENIX Security 2019) shows how these overlap in practice. Of 44 intimate-partner-violence clients, 23 had security vulnerabilities identified, 14 had potential password compromise, 12 had unknown "trusted" devices, and 3 had potential spyware. The authors suggest the low spyware rate may reflect abusers surveilling through compromised accounts, and the fact that almost all clients no longer lived with the abuser and had reset their devices. Of 79 apps their screening tool flagged, 61 were dual-use apps, mostly "find my phone" and child-monitoring apps, and clients recognised all of those but one. The authors note limits: one city, and all but one participant were women and all but one no longer lived with their abuser.
The platform vendors document account-level reviews, none of them a malware scan:
- Apple. Safety Check (iOS 16 or later) lets a user review and stop sharing with people and apps, reset app privacy permissions, and review devices and phone numbers tied to an Apple Account. Apple notes that some options may be unavailable if Screen Time restrictions or an MDM profile are present.
- Google. The devices page lists devices and sessions signed in to a Google Account, and the compromised-account guide covers reviewing security events, unrecognised devices and unusual Location Sharing. Per Google's Location Sharing help, shared information can include real-time location, battery status and recent places, and sharing can be stopped per person. Google also publishes a step-by-step Security Checkup that gives personalised recommendations to strengthen account security. Its support page lists recommendations such as recovery options, passkeys, 2-Step Verification and reviewing linked apps. Neither page describes it as covering location sharing, so this review does not treat it as an equivalent of Apple's Safety Check.
What Android stalkerware abuses
Liu et al. (PoPETS 2023) reverse-engineered 14 leading consumer Android spyware apps (selection used a May 2022 ranking snapshot). They describe the pattern: the abuser needs physical access to install the app, grants it an extensive permission set, "a step which entirely undermines the Android app permission model", and afterwards controls it through a web portal. Findings relevant to triage, described here as capabilities and not as steps:
- Accessibility services. The apps primarily abused the accessibility permission, which needs no root, to read other apps' displayed data and record keystrokes. Android's documentation describes an accessibility service as able to inspect screen content and interact with apps on the user's behalf, intended for assistive tools. Google Play's policy lists monitoring apps among the apps that are not accessibility tools.
- Notification access. Nine of the apps read other apps' notifications, which works even when the spyware is not in the foreground. Android's reference describes the underlying service as one that receives calls from the system when notifications are posted or removed.
- Device administrator. Eleven apps registered as device administrators. The paper says that on Android versions below 7.1 this could disable the Force Stop and Uninstall buttons, while from 7.1 the Uninstall button stays enabled for such apps. The paper adds that this holds on most devices, and that two apps (Cerberus and Mobile-tracker-free) interfered with uninstallation, one by blocking the uninstall screen and requesting an abuser-set password. AV-Comparatives (2025) says stalkerware is often designed to resist removal.
- Hiding. Eleven apps disguised themselves with innocuous names and icons such as "Wi-Fi". The paper also describes hiding the icon by declaring no launcher activity.
Platform and store policy responds in part. Google Play's malware policy prohibits apps from monitoring another individual by collecting and transmitting sensitive data, with an exception for apps designed and marketed for parental or enterprise monitoring that meet requirements including a persistent notification. The AV-Comparatives report (2025) states that most stalkerware apps are sideloaded directly from developer websites, often with instructions to disable built-in protections. Citizen Lab's 2019 technical assessment focused on Android because, it says, Android stalkerware involves installing malware, in contrast to iOS stalkerware, which routinely depends on the target's iCloud password.
Detection resources and their limits
Antivirus. The Coalition publishes detection criteria for vendors, covering apps that market covert spying, can track or control a device without continuous consent, can exfiltrate sensitive data, or hide themselves and resist uninstallation. Its survivor page says good antivirus products should largely detect stalkerware. The EFF and AV-Comparatives test (published 2025-11-06) measured this on one non-rooted Samsung phone running Android 15, with 17 stalkerware apps and 13 products. Malwarebytes detected all 17; Bitdefender, ESET, Kaspersky and McAfee detected 94%; Google Play Protect scored 53% and Trend Micro 59%. EFF calls Play Protect's result unsurprising because some stalkerware includes instructions to disable it during installation. The report also found that most products showed generic alerts, most recommended immediate removal without explaining the risk, and none notified users through a separate channel as the Coalition recommends. The report says in-app and push alerts risk exposing the detection to an abuser monitoring the device in real time. It covers Android only.
Android Intrusion Logging. The Coalition's 2026-10-21 post says this optional addition to Android Advanced Protection Mode keeps logs that cannot be removed, works only once enabled, and is especially useful when a survivor is receiving direct technical support. Its requirements and limits are covered in the MVT guide.
Network-based tools. SpyGuard monitors a device's Wi-Fi traffic against indicators, anomaly rules and Suricata. Its README says it is "not a forensic tool" and may miss malware that does not communicate during the capture, and recommends at least 30 minutes of capture. Stalkerware indicators are managed by ECHAP under a non-commercial licence (CC BY-NC-SA), and the README adds that most digital violence and surveillance uses simple means such as hacking cloud and mail accounts. The Coalition's May 2025 post describes SpyGuard as scanning without alerting the abuser. AV-Comparatives (2021, about its predecessor TinyCheck) cautions that a stalker may still notice a phone connected to an unusual Wi-Fi network. The README has users interact with the phone during capture (reboot, photo, SMS); AV-Comparatives advises doing nothing that would arouse the stalker's suspicion. The Coalition describes TinyCheck as deprecated, and the directory lists it as unmaintained. In the 2021 appendix, AV-Comparatives ran TinyCheck against 20 stalkerware apps on one test smartphone and it detected 15; the report says this should not be compared with its antivirus results, and that only communications sent during the test period are intercepted.
iOS indicator scanning. iMazing Spyware Analyzer says on its vendor page that, as of May 2023, it can detect dozens of different possible stalkerware and watchware products by indicator match, and warns that a match is not proof: visiting mspy.com in Safari alone can raise an alert. It covers iPhones and iPads only.
Forensic tools. The MVT README states it is not intended for end-user self-assessment, and that public indicators cannot show a device is clean. According to its README at v1.8.3, androidqf collects, among other data, a list of installed packages. Both READMEs describe their purpose as consensual analysis of devices of people who might be targets of sophisticated spyware. The androidqf README says its licence permits use only with the explicit consent of the data owner, and the MVT documentation says the same of MVT's modified licence. MVT's public indicator index lists ECHAP's stalkerware indicators, the same managed set SpyGuard uses; the sources reviewed here do not say how much of the commercial stalkerware market they cover.
Where to get help
- Coalition Against Stalkerware. Its resources page lists survivor-support organisations by country, and warns that the person monitoring could see searches for help. The same page notes that a German evidence-documentation app should not be used on a device suspected of being monitored.
- NNEDV Safety Net. The Safety Net Project describes training, technical assistance and resources on technology and abuse.
- Clinic to End Tech Abuse. Cornell Tech's IPV Tech Research group says it runs this clinic and offers free resources to survivors in New York City.
- Access Now Digital Security Helpline. The Helpline serves civil society (activists, media, journalists and human rights defenders) and checks that a request is within its mandate before giving support; its service terms are in the MVT guide.
What is not known
The three Coalition information pages cited here (survivors, tech companies, resources) show no visible date, and their page metadata gives a modified date of 2026-10-16, so the guidance reflects what they said on 2026-11-10. One independent test of a network-based predecessor exists (AV-Comparatives 2021: TinyCheck detected 15 of 20 stalkerware apps on one test smartphone, with the caveats above). No independent detection-rate evaluation of SpyGuard or of iOS stalkerware scanners was found in public sources. The clinic data comes from a single city in 2019, and the app study from 14 Android apps; the market and Android versions have since changed.
Sources
- Coalition Against Stalkerware, Information for survivors (undated, read 2026-11-10)
- Coalition Against Stalkerware, Resources (undated, read 2026-11-10)
- Coalition Against Stalkerware, Information for tech companies (definition and detection criteria; undated, read 2026-11-10)
- Coalition Against Stalkerware, New Android Intrusion Logging Could Help Detect Stalkerware Installations (2026-10-21)
- Coalition Against Stalkerware, SpyGuard vs. Stalkerware (2025-05-19)
- EFF (Eva Galperin), EFF Teams Up With AV Comparatives to Test Android Stalkerware Detection by Major Antivirus Apps (2025-11-06)
- AV-Comparatives with EFF, Stalkerware Test 2025 (test period Aug-Sep 2025, revised 2025-11-03)
- Havron et al., Clinical Computer Security for Victims of Intimate Partner Violence (USENIX Security 2019)
- Liu et al., No Privacy Among Spies: Assessing the Functionality and Insecurity of Consumer Android Spyware Apps (PoPETS 2023, issue 1)
- Citizen Lab, The Predator in Your Pocket (2019-06-12)
- Citizen Lab, The Predator in Your Pocket, full report PDF (2019-06)
- Apple Personal Safety User Guide, Safety Check for an iPhone with iOS 16 or later (read 2026-11-10)
- Google Account Help, See devices with account access (read 2026-11-10)
- Google Account Help, Secure a hacked or compromised Google Account (read 2026-11-10)
- Google Account Help, Manage your Location Sharing settings (read 2026-11-10)
- Google Play Console Help, Malware policy (Stalkerware section; read 2026-11-10)
- Google Play Console Help, Use of the AccessibilityService API (read 2026-11-10)
- Android Developers, Create an accessibility service (read 2026-11-10)
- Android Developers, NotificationListenerService reference (read 2026-11-10)
- SpyGuard README (identical at tag v2.1)
- iMazing Spyware Analyzer (vendor page, read 2026-11-10)
- MVT README (mvt-project/mvt)
- androidqf README at tag v1.8.3
- Access Now Digital Security Helpline (read 2026-11-10)
- MVT documentation, License (read 2026-11-10)
- mvt-project/mvt-indicators, indicators.yaml (read 2026-11-10)
- AV-Comparatives, Android Stalkerware Test 2021 (last revision 2021-05-04), appendix on TinyCheck
- Google Safety Center, Security settings (read 2026-11-10)
- Google Account Help, Make your account more secure (read 2026-11-10)
- NNEDV Safety Net Project, Technology Safety (read 2026-11-10)
- Cornell Tech IPV Tech Research (read 2026-11-10)