1. What we collect
To run the service we store the minimum needed:
- Account data — your email address, name (if you provide it), and a scrambled password hash. We never store your password itself.
- Licenses and devices — your license keys, plan, and the devices you protect (device name, platform, app version, IP address, and last check-in time).
- Security events — detections, quarantined file records (hashes and paths, not contents), and scan history, so your dashboard can show what was found. Since client 2.5.130 a detection report also carries the numeric scores the client computed for the signals that fired (for example an entropy or packing score — numbers only, never file contents), so detection quality can be measured and improved.
- Crash reports — if the Windows client (2.5.130+) crashes, on its next start it sends a crash report: the app version, the exception summary, a short tail of the last actions the app took (action names and timestamps — no file contents, no typed text), and the Windows minidump file for that crash. Reports are capped per device and are used only to find and fix the crash. You can ask us to delete a device's crash reports at any time.
- Settings backups — your devices back up their AEAntivirus settings and file exclusions to your account so a reinstall or a new PC can restore them. This backup holds app settings and exclusion lists only — never passwords, files, or browsing data. You can see when it last synced, and schedule a restore, in My Account → More → Settings backup.
- Home network inventory — a protected PC may report the devices it can see on its own local network (MAC address, local IP address, device name, and hardware vendor) so the admin console can show what is on the network. This list never leaves your account and your servers' admin console.
- Router Guard — the public IP address our server sees on a device's connections is already collected with its check-ins (see Licenses and devices, above). A network's designated main device (a desktop you pick in My Account — one main per network) also reports a summary of its audit of your router: the router's brand and model when discoverable, the gateway address and vendor, the audit score and check results, the port-forwarding count and mappings the router itself reports, the WAN address the router itself reports, and the device's public IPv6 addresses it reports. Router admin passwords never leave the device — where the optional Netgear adapter uses one, it is stored only on your own PC, encrypted — and no router credential is ever sent to or stored on our servers. Separately, a WAN test runs only when you (or an administrator) press its button: it probes the same public IP address our server already sees for your main device's connections — never any other address — at a small set of TCP ports; the IPv6 form of the test probes only the public IPv6 addresses your main device itself reported, under the same button-press rule and the same shared limit of one run per hour, and we store the result (which ports answered, and how they compare with the router's own report) for your account. You can ask us to delete a device's Router Guard reports and WAN test results at any time. For an account with Router Guard active we also keep a history of that main device's audits for each network — the audit score, the check results, the port-opening identities (protocol and external port, with the internal device identified only by a short one-way hash, not its address), the DNS servers the audit observed, and the router's brand, model and firmware — together with a change log derived from them (score changes, check results that flipped, port openings that appeared or disappeared, DNS, firmware or router changes, and WAN test outcomes). The history is kept per network, capped at the newest 100 audits and 90 days, with the change log capped at the newest 300 entries per network. It is visible only to you in My Account (and to our administrators in the admin console, and on your own devices in the app), and it is deleted when your account is deleted.
- Device diagnostics — an administrator can ask a protected device (client 2.5.132+) to run a read-only health check and send the result to the admin console: drive model, serial number, size and health state (including SMART failure prediction) and volume free space; battery design and full-charge capacities; whether the device can reach its gateway, resolve our domain, and connect to our server, with timings; a snapshot of processor, memory, Windows build, uptime and busiest processes; or the verdict of Windows' system-file verification. Client 2.5.133 adds six more read-only checks an admin can request: the programs set to start with Windows (name, command line, where the entry lives, and publisher where reported); motherboard, display adapter and memory-module details; installed Windows updates and the last update results Windows recorded; a grouped digest of critical and error events from the device's System and Application event logs (source, event ID, count and the newest message); the sizes of the system drive's top-level folders and their immediate subfolders (folder names only — never file names below those folders); and a security posture summary (firewall state per profile, UAC and Secure Boot state, BitLocker state per volume, and the antivirus products Windows has registered). Client 2.5.136 adds six more: printer name, status, queued-job count and driver name; the installed-programs list (name, publisher, version, install date and estimated size recorded in the registry); the hardware devices Windows reports a problem code for, with the code in plain words; the automatic services that are not running (name, state and exit code, with totals); the device's local user accounts — this report includes the names of the local user accounts on the device, whether each is disabled, whether a password is required, whether it is an administrator, and its last sign-in time where Windows reports one; and blue-screen history for the last ninety days (the stop code from each bugcheck report where present) with the names, dates and sizes of the crash dump files Windows keeps (never their contents). Client 2.5.137 adds six more: per-disk reliability data (model, serial number, media type, and the wear level, power-on hours, temperature and read/write error totals the drive itself reports — many drives report none of these); network adapters — this report includes MAC addresses and the current network name (SSID), along with each physical adapter\'s IP addresses, gateway, DNS servers, link speed and driver version and date, and the current Wi-Fi signal strength and channel; restore points (creation time, description, type and sequence number) plus shadow-copy storage usage where the device\'s own vssadmin tool reports it; boot performance (the boot durations Windows itself recorded for the last ten boots, and a count of performance-degradation events); activation status (product name, channel and activation state, with grace days where applicable — no product keys are collected: a partial key\'s presence is used on the device only to select the real activation entries, and the key itself is never sent or stored); and USB storage history — this report includes USB storage device serial numbers, with each device\'s name and an approximate last-seen date from the registry (names and serials only — Windows keeps no record there of what files were on a device, and we collect none). Client 2.5.138 adds six more: a system spec sheet (Windows edition, version, build, release and installation date, BIOS details, processor, motherboard — including its serial number, which the hardware inventory already reports — memory, display adapters, and the boot volume and boot drive; the sheet compiles the identity fields the other hardware reports already carry, and its only new items are the BIOS version and date and the Windows installation date); browser extensions — this report includes extension names, versions and extension IDs for Chrome, Edge and Firefox profiles (never extension code, settings or browsing data); a hosts and proxy check — this report includes the hosts file entries verbatim (a hosts file can carry internal network names) and each signed-in user's proxy server and auto-config addresses (accounts that are not signed in are counted as skipped, never forced open); a 90-day history of disk, NTFS and volume errors and warnings from the System log; every scheduled task (path, hidden flag, trigger types and the programs it runs, with Microsoft's own tasks marked); and temperatures from the PC's own thermal zones and drive sensors, where the PC reports any. Client 2.5.139 adds nine more: a driver inventory (device names, driver providers, versions, dates, file names, and whether each driver is digitally signed and by whom); drive encryption and TPM status (whether each drive is encrypted and protected, and whether the TPM chip is present, enabled, and activated, with its versions — the check reads status only and never collects recovery keys, passwords, or any key material); pending-restart state (which Windows servicing flags are set, how many files are queued to be renamed or deleted at the next start and their paths, and a pending computer rename); antivirus products (names of the products Windows has registered, whether each real-time scanner is on, whether its signatures are current, and Windows Defender's state and signature age); app reliability history (names of applications that crashed or hung in the last thirty days, with counts and dates); a memory check (the last memory-diagnostic result, page file names and sizes, and memory use); shared folders (share names, paths, descriptions, types, and connection counts); clock and time sync (the time service state, the configured time source, the last successful synchronization, and the device's own clock reading, which our server compares with its own clock and records the difference); and a performance snapshot (processor and disk load at that moment and the names of the heaviest programs). Client 2.5.140 adds eight more: a remote access inventory (whether Remote Desktop is switched on, its port and sign-in security setting, the Remote Assistance invitation settings, and the names of known remote-access tools found installed or running — names and state only); a firewall rules audit (the names of the enabled inbound allow rules, their protocol, local ports, whether they accept connections from any address or only specific ones, and which firewall profile they belong to); power and sleep settings (the active power plan's name, its sleep, display-off and hibernate timers, USB selective suspend, fast startup, and whether hibernation is available); a browser settings check (each Chrome, Edge and Firefox profile's homepage and startup page addresses and its default search provider — for Chrome and Edge only the provider's name and the host part of its search address are reported, never the full search address; Firefox's search settings are stored compressed on the device and are reported as not read); an environment and PATH check (the PATH entries from the system and from each signed-in user's settings, which of them point at folders that no longer exist, and any duplicates, plus whether the system TEMP folder exists and the core variables are present); saved Wi-Fi networks (the names (SSIDs) of the Wi-Fi networks the device has saved, each network's security type and encryption, and whether it connects automatically — passwords and keys are never read); displays and monitors (each active monitor's manufacturer, model, serial number, year, size and native resolution, and the resolution it is currently running); and backup status (counts of backup starts, successes and failures from the last thirty days of the device's own Windows Backup and File History event history, the last success and failure times, the shadow-copy service states, and per-volume shadow-storage usage). Client 2.5.141 adds the final three checks, completing the catalog at forty-nine: a Windows 11 readiness check — this report includes the Windows edition and build, whether a TPM 2.0 chip is present, whether Secure Boot is on, the processor name, core count and speed, the memory size, and the system drive's total and free space (the readiness verdict itself is computed in the admin console against Microsoft's published minimums — the device sends only these facts, and the processor is reported as speed and cores only, never as a claim about Microsoft's supported-processor list); a cleanup check (the sizes of Windows.old, the Windows Update download cache, the recycle bin on each volume, and the Windows and system temp folders, plus the component store's size on disk — folder sizes only; no file names or contents are read, and nothing is deleted); and a Bluetooth check — this report includes the names and Bluetooth addresses of the devices paired with the PC, a coarse device kind, and a battery level only where Windows itself keeps a cached one for the device (it is never estimated). The battery check also reports the battery's charge-cycle count where the PC reports one. Two things happen with a result once it arrives (server 2.5.179): the plain one-line summary of each check is visible to the device's own account holder under My Account → Billing → Checks by your technician (the check's name, the date, and that summary only — never the full report), and a small set of serious findings alerts us automatically so a technician sees them without waiting for anyone to open the admin panel: a drive reporting bad health, SSD wear of 80% or more, blue screens present, more than one active antivirus or out-of-date antivirus signatures, Windows not activated, or a restart pending. We also keep the one-line summary of each result — plus a few trend numbers where a check has one (such as SSD wear and battery health) — in a per-device history (newest 500 entries per check) so a technician can see how a device changes over time. From server 2.5.185, diagnostics also run on Linux devices (nineteen checks, Linux client 1.27.0 — including drive health read through the machine's own smartctl tool when present, with fields the machine does not report left absent rather than estimated) and on Android devices (ten checks, Android client 1.33.0 — battery charge, health and temperature, storage sizes, connectivity results, device and memory facts, installed application names and versions, security settings state, and the device clock, collected inside the app's existing permissions; Android battery wear is not collected because the system does not expose it; from server 2.5.201 two more read-only audits, still inside the app's existing permissions — the accessibility services enabled on the device (service name and app package) and the apps that can draw over other apps (app name and package)).
The checks change nothing on the device, every run is recorded in the device's own event journal, and we keep the five most recent reports per check per device. You can ask us to delete a device's diagnostic reports at any time.
- Remote fixes — an administrator can also ask a protected device (client 2.5.142+) to carry out one of a short list of fixes (Terms §4t): start a named Windows service; cancel the queued print jobs of one named printer (or, if that method fails, restart the print spooler and clear the shared spool folder, which clears every printer's queue); flush the DNS resolver cache; resync the device's clock; delete the contents of chosen junk locations (the Windows and system temp folders, the Windows Update download cache, and the recycle bins); delete Windows.old (which permanently removes the option to roll back the last Windows upgrade); reset the Windows Update components (the update downloads folder is renamed aside and kept for 7 days, then deleted); undo that reset within the 7 days; or run the system-file repair scan. Client 2.5.143 adds six more: disable one startup entry (a registry Run/RunOnce value is moved — not deleted — into the product's own holding key on the device under the same registry hive, keeping its name and contents; a Startup-folder item is renamed in place with a ".ae-disabled" suffix), re-enable a held startup entry, remove one PATH entry (only after the device verifies the entry is in the device's system PATH or a signed-in user's PATH and that the folder it points to does not exist; the full original PATH value is first backed up on the device itself, in the product's own registry backup area, and can be written back by the restore fix, which marks the backup used), remove one network share (only the share is removed — the folder and its files are never touched or read; Windows' own standard shares are refused, and there is no undo), and run the DISM image repair (DISM /Online /Cleanup-Image /RestoreHealth, run with a fixed path and fixed arguments exactly like the system-file repair scan). The PATH backups and held startup entries stay on the device; a fix outcome report names what was done (for a PATH removal, which PATH was edited, how many occurrences were removed, and the backup's timestamp id; for a share removal, the share name and the folder path it pointed to) — never the contents of any file or folder. A fix command can only ever choose from that short list of known actions — it cannot run programs or open files on the device: the only things a fix can name are a service name, a printer name, or the fixed cleanup locations above, and the device verifies the target against its own services and printers before acting. A fix sends us an outcome report (what was done, whether it worked, and measurements such as the space freed and, for the clock fix, the device's clock reading, which our server compares with its own clock) — never file contents: the cleanup deletes files without reading or sending them, and no report lists the files it deleted. While a fix runs, the device shows a banner saying a technician is applying a fix. Every fix run is recorded in the device's own event journal before it starts, and we keep the five most recent outcome reports per fix per device. You can ask us to delete a device's fix reports at any time.
- Remote terminal sessions — the shop's owner (only the owner — not manager or support staff) can also ask a protected device (client 2.5.144+) to open a remote terminal session (Terms §4u). A terminal session starts only when the person at the device approves it: the consent prompt on the device names the business, explains that the technician can type commands and see the results, states that the session runs with full system access, and states that it is recorded; it declines itself after 60 seconds with no answer. A session is recorded as a transcript — the commands typed and the output returned, kept with the device's service history and accessible to the shop's owner only; stored output is capped at 1 MB per session, and the transcript records when the cap was reached. Transcripts are never exposed on customer account pages. During a session the device shows a persistent banner with an End session control, and sessions end automatically after 15 idle minutes or 2 hours. Within a session the technician may open a tunnel to the device's own web interfaces only (the device's loopback address, ports 80, 443, 8080, 8443): the tunnel's existence and the number of bytes proxied are recorded, but the content of the pages proxied is not stored. Within a session the technician may also open a raw tunnel to the device's own Remote Desktop (3389), VNC (5900), or SSH (22) service — again the device's loopback address only, and only those three ports — through a companion program on the technician's own computer (client 2.5.145+): the tunnel's existence, its port, its open and close times, and the number of bytes relayed are recorded, but the content of a raw tunnel is never recorded or stored. You can ask us to delete a device's terminal transcripts at any time. A terminal session is never billed on its own; if the shop records a session as a repair charge afterwards, the charge appears in your account's service-fee history and is billed under the same terms as other repairs.
- Cloud storage files — files you choose to upload. They are private to your account and are never scanned for advertising, shared, or sold.
- Support messages — what you send through the contact form, so we can answer you. Since server 2.5.191 that also covers a checkup request you make from My Account: the device it is for, your optional short note, and the request's status and dates, so your technician can act on it.
- Checkup reviews — if you leave a review after a technician checkup (server 2.5.193), we store the review text and the display name you choose for it. A review is shown publicly (on our Compare page) only when you tick the "You may show this review on aeantivirus.com" box and we approve it; without that permission it is private feedback that we can read but never display. You can edit or delete your review at any time from My Account, and editing a shown review takes it down until it is approved again.
- Payment data — handled by Stripe, our payment processor. We see the last four digits and expiry of your card at most; full card numbers never touch our servers.
2. What we do with it
Your data is used to provide the service: protect your devices, bill your plan, show your dashboard, and support you. We do not sell your personal data, and we do not use it for advertising.
3. What we share
- Stripe — to process payments (subject to Stripe's privacy policy).
- Threat-intelligence lookups — when the app checks a file hash against threat databases, only the hash is sent, never the file.
- Webhook endpoints you configure — if you register a webhook (Business, Enterprise and Custom plans), the events you subscribe to (confirmed detections, license expiry warnings, device-offline notices) are sent to that endpoint. This sharing happens only because you set it up, and deleting the webhook stops it.
- Your administrator's security systems (optional, off by default) — on business deployments, the service operator can turn on syslog/SIEM forwarding, which sends detection events, tamper events and (optionally) admin audit entries to a security collector the operator themselves designates. Events contain device identifiers, threat names and timestamps — never file contents or browsing history. This is configured by your own administrator, for your own organisation's security monitoring.
- Improving the threat model — when you or your administrator mark a detection as a real threat or a false positive, the verdict is recorded with the technical signals that fired (for example "high-entropy" or "injection-imports") and the timestamp. Those signal tags and verdicts — never file contents, file names beyond the flagged path already stored with the detection, or any browsing data — are used to train the on-server threat model that makes future detections more accurate. Verdicts you mark stay associated with your fleet's detections.
- Legal — if required by law, we will disclose the minimum necessary and tell you unless we are forbidden to.
We do not share your data with data brokers or advertisers.
4. Cloud storage privacy
Files you store with the cloud storage add-on are encrypted in transit and stored on our servers, visible only to your account. Our staff does not browse your files. A file confirmed malicious by our threat engine may be quarantined from storage and you will be notified.
5. Cookies
The site uses strictly necessary cookies only: a session cookie to keep you signed in, and a cookie to remember security preferences. There are no advertising or cross-site tracking cookies, and we do not use third-party analytics that follow you around the web.
6. Data retention
Account and license records are kept while your account exists. Security event history is kept for up to 12 months, then pruned. Cloud storage files are kept until you delete them. If you close your account, contact us and we will delete your personal data within 30 days, except records we must keep for tax or legal reasons.
7. Your rights
You can download your data anytime from your account page (licenses, devices, and sessions as one JSON file). You can correct your name and email there too. To delete your account or ask what we hold about you, contact us.
8. Security
Passwords are stored as salted hashes, sessions expire, and admin access requires two-factor authentication. No system is perfect — if we discover a breach affecting your data, we will tell you promptly.
9. Changes
If this policy changes in a meaningful way, we will note the new date above and, for significant changes, notify you by email.
10. Contact
Privacy questions: use the contact form and mention "privacy".