Detection & Protection Disclaimer
GreyArm identifies known and detectable AI activity. It cannot guarantee to detect, stop or contain every AI agent. Please read this.
Key points
- 01GreyArm identifies AI agents, models and services that are known to us or detectable with the signals available on your devices and network. It does not and cannot detect everything.
- 02New AI models, agents and services appear constantly, and some are designed to be hard to spot. There will always be AI activity GreyArm does not recognize.
- 03Containment and blocking are best-effort. GreyArm does not guarantee it can stop, block, remove or contain any AI agent or prevent any outcome.
- 04Risk scores, labels and alerts are automated estimates that can be wrong — including false alarms and missed activity. Use your own judgment before acting on them.
- 05GreyArm is not a substitute for antivirus, a firewall, backups, parental supervision or professional security advice. You remain responsible for your devices, accounts and data.
This summary is for convenience. The full text below is what governs.
01Why this document exists
GreyArm, LLC (“GreyArm”, “we”, “us”) builds software that helps people see which artificial-intelligence agents, assistants, models and services are running on, or communicating from, the devices and networks they control. We want you to understand exactly what that software can and cannot do before you rely on it.
This Detection & Protection Disclaimer (“Disclaimer”) forms part of, and is incorporated into, our Terms of Service and End User License Agreement. By installing or using any GreyArm software or service you acknowledge that you have read and accept it. If you do not accept it, do not install or use GreyArm.
GreyArm is a detection and visibility tool for known and detectable sources of AI activity. It is not, and is not sold as, a guarantee that all AI activity will be found, identified, stopped or prevented.
02How detection works, and where it stops
GreyArm’s findings come from a catalog that we maintain and from rules that look at observable signals, such as:
- the names, file locations, command lines and code-signing publishers of running programs;
- network connections from your devices to internet addresses and domain names associated with AI providers;
- domain-name lookups seen by the optional DNS shield; and
- behavior patterns that our rules associate with automated or AI-driven software.
Each of these signals has limits. As a result, GreyArm may fail to detect, or may misidentify, AI activity that:
- uses a model, agent, tool, provider or version that is new, private, custom-built or not yet in our catalog;
- is renamed, repackaged, obfuscated, encrypted, tunneled, proxied, routed through a VPN, relay or general-purpose cloud service, or otherwise disguised;
- runs entirely offline or inside a browser tab, virtual machine, container, remote desktop, cloud workspace or another device we cannot observe;
- runs on a device where GreyArm is not installed, is not running, is out of date, lacks the permissions it needs, or has been disabled or tampered with;
- is embedded inside an ordinary application, website, operating-system feature, keyboard, game, toy or smart-home product; or
- communicates over channels the DNS shield does not see, including encrypted DNS, hard-coded addresses, private relays or mobile networks.
The pace of change in artificial intelligence means our catalog will always lag some part of what exists. We update it regularly, but no update schedule can keep up with every new release.
03Containment, blocking and the DNS shield
Some GreyArm features let you act on what you find — for example, ending a process, adding firewall rules, or refusing DNS lookups for AI services. These actions depend on your operating system, your network equipment, the permissions you grant, and the behavior of the software being acted on. They may fail, partially succeed, be reversed by the software or the operating system, or be bypassed.
Containment and blocking can also have side effects, including stopping software you rely on, interrupting work in progress, breaking features of applications or websites, and preventing legitimate services from connecting. You choose when to use these features and you are responsible for the result.
We do not guarantee that any containment, block or other action will stop, remove, disable or prevent any AI agent, model, service or activity, or will do so without side effects.
04Risk scores, labels and alerts
Risk levels, scores, categories, names, vendor attributions, explanations and alerts are produced automatically from the signals described above. They are estimates, not findings of fact, and they can be incomplete or wrong. A “low risk” label does not mean software is safe, and a “high risk” label does not mean software is malicious.
Alerts may be delayed or not delivered at all — for example, if a device is offline, notifications are disabled, an email is filtered, or a service we rely on is unavailable.
05What GreyArm is not
GreyArm is not antivirus or anti-malware software, a firewall, an intrusion-prevention system, a backup product, a data-loss-prevention product, a parental-control or content-filtering service, or a managed security service. It does not provide legal, compliance, security-consulting or parenting advice.
We recommend that you keep your operating systems and applications updated, use reputable security software, keep current backups of anything you cannot afford to lose, use strong and unique passwords with multi-factor authentication, and supervise children’s use of technology directly.
06Your responsibility
You are solely responsible for:
- deciding whether GreyArm is appropriate for your needs;
- the security of your devices, networks, accounts, credentials and data;
- reviewing alerts and findings before acting on them, and any action you take or decide not to take;
- making sure you are permitted to monitor every device and network on which you use GreyArm (see our Acceptable Use Policy); and
- complying with laws, regulations and agreements (including employer, school or software-license terms) that apply to you.
07No guarantee of detection or protection
08Limitation of liability
The limitations of liability, disclaimers of warranty, indemnity and dispute-resolution provisions in our Terms of Service and End User License Agreement apply in full to any claim connected with detection, non-detection, misidentification, alerts, containment, blocking or the DNS shield.
Some jurisdictions do not allow certain warranties to be disclaimed or liability to be limited. In those jurisdictions, our disclaimers and limitations apply to the greatest extent the law allows, and nothing in this Disclaimer limits rights you have that cannot lawfully be waived.
09Changes and contact
We may update this Disclaimer as the product changes. The effective date above shows when it last changed, and material changes will be announced in the product or by email. Questions: support@greyarm.com.