The conversation about therapists using AI in therapy often skips straight to the gadgets like transcription, note-taking, scheduling, and the efficiencies they offer. But beyond technology, we need to talk about consent: the kind that isn’t merely a signature on a form, but an active, ongoing process. 

Clients may feel uncertain or hesitant about AI, and your role is to offer clear information and invite them to make an informed choice without pressure. However, it can feel like you need to become an AI expert to have these conversations around consent. Here’s clear steps about how to guide that conversation in a way that gives your clients the fullest possible agency – and with language you can actually use in the room.

1. Start with Clear, Informed Consent

Consent is more than a signature. In their therapy sessions where the stakes are emotional safety and personal history, it matters that your client actually understands the tool and feels comfortable using it. This is why discussing AI and providing information about a few key topics is necessary. Keep in mind that how you say something is just as important as what you are saying. 

First, you’ll want to obtain clear, informed consent. This can sound like:

“I use an AI transcription tool in some sessions to help me take accurate notes. It temporarily records our conversation to automatically generate a progress note, which is a required form of clinical documentation. You don’t have to agree to this, we can do our work without it. If you’d like, I can explain how it works and what happens with the recording.”

• If client agrees: “Great. I’ll explain a little more how it works, and check in with you once in a while in case you have questions or want to change your mind.”
• If client declines:“That’s perfectly okay. Our work together won’t change.”

2. Explain Data Security in Plain Language 

Second, clients need to know how their information is stored and protected. Simple explanations go a long way to spell out what that means in practice. It’s also worth sharing which specific security measures are in place to prevent unauthorized access. For example, if you’re using a tool like Blueprint, you can emphasize to your client that the tool is HIPAA compliant, and is SOC 2 Type II certified, which means it goes through an external audit every year. This can sound something like: 

“The transcription file is encrypted and HIPAA compliant, which means no one without permission can read it. It’s stored on Blueprint’s system, which is audited every year to ensure SOC 2 Type II data privacy and security certification standards.”

• If client agrees: “Wonderful. I’ll update you if anything changes with the program or the way they handle data.”
• If client declines or expresses concern: “I understand. We can keep things manual without the AI.”

3. Be Transparent About What Happens to Their Data

Last, spell out exactly what AI has access to and for what purpose. If it’s a recording of a session, say how long that recording exists before deletion. If it’s a transcription, explain its retention schedule, and any other relevant details. For instance, if you’re using Blueprint, this may sound like: 

“The AI only processes the recording of our session. It doesn’t connect to your medical records or anything else. The recording is deleted immediately after the progress note is generated, and I can permanently delete the transcription at any time. This information will never be used to train AI models.”

• If client agrees: “Thanks for your trust. I’m happy to remind you about session recordings or transcripts being deleted if you want.”
• If client declines: “No problem at all. We’ll continue without it.”

4. Invite Open Choice Without Pressure

As mentioned earlier, your approach and language is just as important as the information you are providing. Consent is an ongoing collaboration, not compliance, so make sure your client knows they can say no or change their mind at any time without it affecting their care. Encouraging that sense of agency may sound something like:

“If you’d prefer not to use AI, that’s absolutely fine. We’ll keep working the way we have been. And if you change your mind later, just let me know.”

• If client agrees: “Thank you for letting me know your preference.”
• If client declines: “I appreciate your honesty, and we’ll go at your pace.”

5. Frame AI as a Support, Not a Replacement

The fear for many clients is that AI will replace the human connection they came to therapy for. Make it clear that AI helps you focus on the session, and does not replace your role – in fact, AI is there so you can be more present, not less. If you’re using a tool like Blueprint AI Notetaker, you can say something like:

“The AI tool helps me take notes so I can stay more focused on you during sessions, capture details accurately, and see patterns I might otherwise miss. But I’m the one reviewing everything and making decisions about your care.”

• If client agrees: “Great. With this tool, I can give you my full attention.”
• If client declines: “That’s completely fine, we’ll continue as usual.”

Consent is an Ongoing Conversation

You have the choice to shape how this technology enters your practice, and using AI tools in therapy should be a decision you, and your client feels comfortable making. Consent isn’t a box checked at intake; it’s a conversation you keep having. Your client’s privacy and trust will always remain your priority, regardless of the technology you use. If you want a starting point for an informed consent document, feel free to make Blueprint’s version your own.