AI Is Already in Your Therapy Practice. Is Your Ethics Policy?

You finish a late session and let documentation software draft the note while you grab a glass of water. A new client fills out an intake form, and an automated summary lands in your inbox before you've even read their actual answers. Reminders go out to clients through scheduling software you set up months ago and barely think about anymore. 

None of this feels like a big decision. It feels like a normal Tuesday. But each of these small moments touches client information, and each one raises real AI ethics therapy questions that most practices haven't actually sat down and answered. The tools are already here. The policy usually isn't.

It's Probably Closer Than You Think

Most clinicians picture chatbots or clinical decision tools when they imagine AI in mental health care. In practice, it tends to show up earlier and quieter than that. Progress notes, appointment reminders, intake processing, even a quick email draft, these are the ordinary corners where automated tools have already found a place. None of this is inherently a problem. Software helping with paperwork isn't new. What's changed is how much client information now passes through systems nobody has formally reviewed and how easy it becomes to stop noticing.

A Five Minute Practice Check

Before anything else, take an honest look at what's actually running in your practice.

• Documentation: Does anything generate or edit your progress notes?
• Intake: Are client responses summarized or processed automatically?
• Scheduling: Do reminders or messages go out without your direct review?
• Storage: Where does that information sit once the task is done?
• Access: Who else, including a vendor, could see it?

You don't need a technical audit for this. You need a plain list. The real question was never which software your practice happens to use. It's what information enters it and what happens to that information once the task is finished.

Convenience Doesn't Reduce Responsibility

Here's the part that gets missed most often. A task can feel purely administrative while it's still touching confidential information, clinical judgment, and a client's trust in you. Handing part of a job to software doesn't hand off the professional responsibility that comes with it. 

If a documentation tool drafts a note that overstates a symptom, smooths over nuance, or adds a conclusion you never reached, that note still carries your name and your license. Fast is only useful when the finished record still reflects your own clinical thinking, not a polished guess at what you meant.

What Clients Actually Deserve to Know

This isn't a legal form to hand over. It's a conversation clients should be able to follow. Before any tool touches their care, consider whether they know:

• Which tool is involved, in plain terms
• What information it processes or stores
• Any privacy or confidentiality tradeoffs
• Whether they have the option to decline
• What alternative exists if they say no

A client doesn't need a technical breakdown. They need enough information to make a real choice, not a line they signed without reading it.

Put the Rules in Writing Before You Need Them

This is where a written therapist AI policy earns its place. Not as paperwork you draft after something goes wrong, but as a set of decisions made ahead of time. Which tools are acceptable. What client information can and can't be entered. Who reviews outputs before they become part of a record. What happens if a tool fails or a mistake slips through. Deciding these questions in a calm moment beats improvising an answer during a difficult one.

What Belongs in That Policy

A working policy doesn't need to be long. It just needs to cover the basics.

• Approved and restricted tools
• What client information can and cannot be entered
• Consent steps for clients
• Who reviews outputs before they're finalized
• Account and access security
• What happens if something goes wrong

Write it once, revisit it yearly, and it stops being a document nobody remembers exists.

Your Policy Should Catch Up With Your Practice

None of this means rejecting every tool that touches your day. It means knowing what's running, what it touches, and where your responsibility as a clinician still sits, no matter how the note actually got written. California's law and ethics CE California requirements exist for exactly this kind of gap between everyday practice and written policy. 

Rouse Academy's 3 CE course, Law & Ethics of AI in Mental Health Care: Privacy, Security, and Live Practice, walks through these exact questions and counts toward your required Law and Ethics hours, giving you a practical way to close that gap instead of guessing at it.