Security and compliance

The people who reach out trust you. We protect that trust.

Someone asking about treatment is sharing something private, often for the first time. Tars is HIPAA compliant, SOC 2 Type 2 audited, and signs a BAA with your center before any patient data moves. This page is written for your compliance, legal and IT reviewers.

At a glance

Built for the rules your work runs on.

HIPAA compliance badge
HIPAA compliantAccess-controlled, BAA in place
SOC 2 Type 2 badge
SOC 2 Type 2Independently audited
BUSINESS ASSOCIATE AGREEMENT •BAA
BAA signedBefore any patient data moves
CRISIS PROTOCOL • 988 • 911 •988
Crisis protocol988 and 911, configured per center
Compliance and audits

Checked by outside auditors, not just by us.

HIPAA compliant

Tars is built to handle protected health information under HIPAA. We protect the conversations people have with your agent as patient data.

SOC 2 Type 2

Our security controls are independently audited over time, not checked once. The SOC 2 Type 2 report is available to your team on request.

VAPT and other security audits

Vulnerability assessment and penetration testing (VAPT), both internal and external, plus other security audit reports. Available on request for your review.

BAA signed first

We sign a Business Associate Agreement with your center before any patient data moves.

Your data

Where it lives, who sees it, and what it's never used for.

Stored in the US

Conversation data is stored in the United States.

Access controlled and limited

Inside your center, only the staff you give access to can see conversations. At Tars, access is controlled and limited to the people who support your account.

Not used to train AI models

Your conversations are not used to train AI models.

Never sent to ad platforms

Conversation data is never shared with any advertising platform. No retargeting pixels fed with what people tell your agent.

Retention you control

How long conversation data is kept is configurable for each center, to match your own policies.

Delete any time

Your center can delete its data at any time.

Safe conversations

Operational, never clinical.

Security is also about what the agent says. These rules hold on chat and on the phone.

AI disclosure by default

The chat agent says it's a virtual assistant, and the voice agent says it's an AI at the start of each call. People can ask for a person at any time.

No clinical advice

It never diagnoses, recommends treatment, or promises a bed, an appointment, coverage or outcomes. It answers only from content your team approves. Care decisions stay with your staff.

Your crisis protocol

When someone may be in danger, the agent shares 988 and 911, flags the conversation and can alert on-call staff, as configured for your center. The voice agent doesn't handle a crisis: it detects it, tells the caller to call 988 or 911, and can transfer to staff if someone is available.

Human takeover

Your staff can watch any chat as it happens and take over with one click. Calls can be transferred live to on-call staff.

Staff-only access to conversations

Conversations and summaries go to the people and systems you choose: your CRM or EHR, and alerts to named staff. Not to anyone else.

Reviewed before go-live

Your clinical, compliance and marketing leads see what the agent will and won't say, and your team tests it end to end before it talks to anyone.

Need our SOC 2 report, VAPT reports or BAA?

Tell us what your compliance team needs to review, and we'll send the documents.

Questions

What compliance and legal reviewers ask us

Will you sign a BAA?

Yes. We sign a Business Associate Agreement with your center before any patient data moves.

Can we see your SOC 2 Type 2 report?

Yes. Our SOC 2 Type 2 report, VAPT reports (internal and external) and other security audit reports are available on request. Contact us to request them.

Where is our data stored?

In the United States.

Who can see our conversations?

On your side, only the staff you give access to. At Tars, access is controlled and limited to the people who support your account.

Is our data used to train AI models?

No. Conversation data is not used to train AI models, and it's never sent to ad platforms.

How long do you keep conversation data?

Retention is configurable for each center, so you can match your own policies. Your center can delete its data at any time.

Does the agent give medical or clinical advice?

No. It's an operational intake agent. It answers from content you approve and passes anything clinical to your team.

How does it handle someone in crisis?

Your crisis protocol runs first: it shares 988 and 911, flags the conversation and can alert on-call staff. You set the exact steps and wording during setup. The voice agent detects a crisis and points the caller to 988 or 911, and can transfer to staff if available; it does not handle the crisis itself.

Do people know they're talking to an AI?

Yes, by default. Your center can adjust how the disclosure is shown, and people can ask for a person at any time.

Do you support 42 CFR Part 2 programs?

Talk to us about your program's requirements. We'll walk your compliance team through how Tars handles data so you can make that call with full information.

Make sure the next person who reaches out hears back.

See Tars answer an inquiry for your center in a 30-minute demo, or get a free chat agent built from your website.

Book a demo

HIPAA compliant · BAA signed · Built only for behavioral health