Bitdefender GravityZone AI Visibility and Control
Your staff already use AI tools. AI Visibility and Control tells you which ones are running on the machines you already manage and what exposure each one creates; PHASR cuts each user and each agent down to what they need. Both ride on the GravityZone agent you already have.
We sell the licence. Deployment is handled by your IT team or your usual partner.
The problem is not AI, it is not knowing which AI is in use
Staff already use AI tools, and most organisations have no reliable way to see which ones are running, or to classify and reduce the exposure each one brings. You cannot govern what you cannot see.
The two products
They are licensed separately and each stands on its own, but they fit together: one shows you what AI is there, the other limits what it can do.
GravityZone AI Visibility and Control
Finds and ranks the AI already in use
Volume discounts
See how it worksGravityZone PHASR
Cuts each user and each agent down to what they need
Volume discounts
See what PHASR doesPrice per seat per year, VAT not included, paid annually, with a 5-seat minimum. Above that there are volume discounts: tell us how many seats you need and we will apply the band to the offer.
It needs GravityZone underneath
Both are add-ons to the GravityZone platform, not standalone products. They work through the agents you already have deployed as soon as the licence is activated, so a GravityZone Business Security licence comes first.
shieldSee GravityZone Business SecurityWhat it gives you
AI adoption without data leaks
A central inventory discovers the AI in use on the endpoints you already manage and helps you identify shadow AI, so you know what is there before deciding what to do about it.
Reduce the AI risks that matter
Findings are grouped by risk category and ranked by severity, each tied to the users and endpoints involved, so a small team acts on what actually matters.
Govern adoption without adding complexity
You act on that evidence using the web filtering, application control and hardening controls you already have in your GravityZone console, without running a separate tool.
How it works
1. Discover
Discover the AI in use through the GravityZone agent already on your endpoints.
- checkLLMs and chatbots
- checkAI assistants
- checkAI agents
- checkMCP servers
- checkTools and capabilities
- checkCoding assistants
- checkSaaS services with AI
2. Assess
See which usage creates real exposure and what deserves your attention first.
- checkFindings ranked by severity and priority
- checkFindings tied to the affected users and endpoints
- checkInformation on the most recent detection
- checkSafe evidence, with no prompt content collected
3. Act
Act on that evidence using the controls you already have in your console.
- checkWeb access control
- checkApplication control
- checkEndpoint hardening policy
- checkBehaviour-based agent hardening, with PHASR
PHASR: hardening tailored to each user
PHASR profiles what each user actually does with their tools and takes away what they do not need. It is built for the attacks that break nothing: the intruder logs in with stolen credentials and moves using the administrative tools already on the machine, such as PowerShell or WMI, which antivirus cannot block without stopping legitimate work too.
- check_circle It is not an allow list: it restricts risky actions inside tools that remain authorised, rather than blocking the whole tool.
- check_circle Two operating modes: Autopilot, which applies the recommendations on its own, and Direct Control, if you would rather approve them one by one.
- check_circle Runs on Windows, macOS and Linux.
- check_circle The first recommendations appear within 30 minutes of switching it on, because it uses the agent already deployed.
- check_circle With AI Visibility and Control alongside it, hardening tells the person apart from the AI agent running on their machine: the agent keeps PowerShell if it needs it and the person does not.
Bitdefender puts living-off-the-land techniques in 84% of major attacks, and around 70% attack surface reduction at one of its customers. Those are the vendor's own analysis and testimonial, not an independent third party, and we present them as such.
What makes it different
- check_circle Discovery runs from the GravityZone agent already on your endpoints, and picks up AI tools that would not be visible at network level, such as local models.
- check_circle Visibility is broad: public chatbots and LLMs, local models, agents, MCP servers, coding assistants and SaaS services.
- check_circle Each finding shows severity and priority, service category, affected users and endpoints, origin, confidence level, and first and last detection dates.
- check_circle Evidence is safe by design: a finding carries what you need to decide, without collecting prompt content or what your staff type into AI tools.
- check_circle With a PHASR licence, hardening treats an AI agent as an autonomous actor: an agent that needs PowerShell can keep it while the human on the same machine cannot, so PHASR restricts the person without breaking the agent.
Why does discovery start at the endpoint?
Most AI usage never crosses a network gateway in any recognisable way. Local runtimes, in-editor coding assistants and MCP servers live on the endpoint, which is where GravityZone is already running.
Securing AI use without blocking it by default
Blanket restrictions stop useful work and get worked around anyway. GravityZone restricts only what creates real risk, and with a PHASR licence autonomous hardening cuts each user and each agent down to what they actually need.
Frequently asked questions
Do I need Bitdefender already to buy it? expand_more
Does it read what my staff type into ChatGPT? expand_more
Does it detect models running locally, with no internet traffic? expand_more
What can I do when I find something I do not want? expand_more
What does PHASR add here? expand_more
Can I pay monthly? expand_more
Related services
How many seats do you need to protect?
Tell us how many machines and whether you have your own IT team, and we will tell you which of the three fits. No commitment.