AI: Oregon SB 1546 Signed Into Law

Oregon Governor Tina Kotek signed Senate Bill 1546 into law on April 1, 2026. This is Oregon’s first attempt to address risks associated with AI chatbots. It aims to protect minors, take precautions for suicidal ideation and ensure individuals know they are communicating with a chat bot.

In December, President Trump issued an Executive Order that set up an AI Litigation Task Force that would challenge state laws, claiming that a patchwork of laws would make it hard for businesses. To date, however, the federal government has not enacted legislation aimed at limiting AI-related harm, protecting licensed professions or establishing enforceable consumer safeguards.

As a result, states are moving forward with their own regulatory approaches.

This Oregon law has five aspects:

Disclosure requirements (A-level regulation)*

The law requires chatbots to clearly disclose that they are not human. This aligns with legislation in other states, including California (SB 243) and Nevada (HB 2225), which focus on representation and transparency.

Suicide and self-harm intervention (C-level regulation)*

AI chatbots are required to:

  • detect suicidal ideation and self-harm signals
  • provide referrals to 988 and other crisis resources

This is a meaningful shift. It treats chatbots as systems that must identify and respond to risk, not simply provide information.

Protections for minors

The law includes several provisions designed to reduce compulsive or harmful engagement:

  • suggesting breaks after extended use (every three hours)
  • prohibiting sexually explicit content
  • banning emotional manipulation tactics designed to prolong engagement
  • restricting reward loops that encourage continued interaction

The restriction of reward loops is particularly significant. It represents an early attempt to regulate what can be described as engagement-driven reinforcement mechanisms in AI systems. This has been a particular concern of mine for all users, not just minors and the emotional/mentally vulnerable.

Reporting requirements

AI systems must file annual reports that include:

  • the number of crisis referrals (988)
  • a description of intervention protocols

This introduces a form of system-level visibility into how often high-risk interactions occur and how they are handled. We’ll see if AI systems actually follow this part of the law.

Private right of action

The law allows individuals to sue for $1,000 per violation. This creates direct liability exposure and removes the possibility of relying solely on internal compliance or voluntary standards. Allowing individuals to sue prevents the AI systems from having a shield law. Significant corporate behavior and future legislation will certainly be informed by court cases where plaintiffs allege harm was done.

Significance of this law

Oregon is moving beyond regulating what AI systems say and beginning to regulate what they are required to do in situations involving risk.

That is a meaningful shift. It places AI systems closer to functioning as actors within a clinical or quasi-clinical space, with corresponding expectations around intervention and accountability.

How companies respond, both in compliance and in litigation, will likely shape the next phase of state and federal AI policy.

A/B/C/D Framework for AI in Clinical and Professional Practice

To make sense of emerging AI laws and policies, I classify how AI is being used and regulated across four categories:

A — Representation and Communication

AI systems that:

  • interact with users
  • provide information
  • simulate conversation

Regulatory focus:

  • disclosure (AI vs human)
  • prohibition on impersonating licensed professionals
  • consumer protection

Example: laws requiring chatbots to identify themselves as AI

B — Structured Intake and Screening

AI systems that:

  • gather user information
  • ask structured questions
  • assist in initial assessment

Regulatory focus:

  • accuracy of screening
  • escalation protocols
  • documentation and oversight

Key issue: missed risk signals or incomplete assessment

C — Triage and Gatekeeping

AI systems that:

  • prioritize cases
  • determine urgency
  • decide whether and how users receive care or intervention

Regulatory focus:

  • duty to escalate (suicidal ideation)
  • limits on autonomous decision-making
  • liability for delayed or missed intervention

Key issue: harm caused by failure to act or improper prioritization

D — Clinical Decision Support (Post-Contact)

AI systems that:

  • assist licensed professionals
  • provide recommendations within ongoing care

Regulatory focus:

  • clinician accountability
  • documentation of AI use
  • maintaining professional standards

Key issue: overreliance on AI without independent verification

Monopoly Money and Moral Conviction: Teaching Critical Thinking Through Scarcity

Each semester, I hand every student $686 in Monopoly money. Five to eight students come to the front of the room. Each has two minutes to propose a federal or state policy: What does it cost? What is the intended benefit? Who does it affect? Who supports it? Who opposes it? Students have presented on abortion policy, climate initiatives, foster care reform, student loan forgiveness and affordable housing. The range reflects student interests and varies each semester.

After the presentations, students privately allocate their $686 to whichever proposals they wish. They may divide it or place it all in one proposal. We total the funds and rank the proposals from most supported to least. Then we analyze the results.

I begin my policy class with a definition from Stuart Shapiro at Rutgers’ Bloustein School: public policy is the allocation of finite public resources. In an era of misinformation and AI-generated fluency, disciplined prioritization matters more than volume.

The exercise is intentionally simple. I do not allow students to spend money to oppose a proposal. I do not allow bargaining or horse-trading. This is not a legislative simulation. It is a priority test. The goal is clarity.

Monopoly money works because people know bankruptcy, overextension and aggressive acquisition from the game. We take that structure and apply it to public budgeting.

The most important moment is not the ranking. It is the debrief. In one class composed mostly of young women, abortion policy ranked near the bottom. That result unsettled assumptions. Cultural intensity shifted when students faced finite allocation. One student later told me, “I’ve been a far left progressive activist since I’ve been aware. I now see that some of my views are impossible to pass as legislation and actually lose me other support.” That movement, from conviction to feasibility, is critical thinking.

The point is not a single ranking. Students repeatedly confront limits, defend claims and revise their reasoning until clarity becomes expectation rather than exception.

Social work education rightly emphasizes dignity and justice. Professional ethics and constitutional protections are not subject to allocation here; the exercise examines discretionary funding priorities above that floor. What students practice is something different: translating conviction into policy under fiscal and political constraint. Social workers advocate, testify and operate within constrained systems. They must distinguish between moral belief and legislative viability.

Preparing them only for the ideal world leaves them unprepared for the systems they will enter.

By the end of the semester, students anticipate scrutiny. They expect to justify cost, anticipate opposition and clarify claims before being asked. That anticipation is the habit.

Critical thinking is not simply identifying bias or critiquing sources. It is ranking priorities under constraint, articulating those priorities clearly and revising in light of feasibility. Monopoly money makes the constraint visible. Precision makes judgment more reliable under pressure.

The Proportional Distress Scale

From the Greenagel Equations

The Greenagel Equations are a set of practical frameworks developed between 2005 and 2008 in schools, outpatient and family treatment settings. They were built in rooms, not in theory, and have been used with students, families, law enforcement, veterans and therapists.

My 11 year-old niece was crying when I showed up. Flowing tears, heaving breaths and red-faced. I asked her what was wrong. It was hard to understand her, but she told me that her younger brother had ripped one of the eyes off of her stuffed cow.

“I don’t like it when someone breaks my stuff either. You have a right to be upset. I’m only here for a little bit though and I’d like to see you. How much longer do you want to cry for? An hour? Twenty minutes? Ten minutes? The rest of the night?”

Her crying almost stopped. “Ten minutes.”

“Ok. When you are ready, I’ll teach you a little trick to deal with stress.”

She changed her mind. “Two minutes.” She had stopped crying.

I smiled.

“I’m ready.”

“What was the worst moment of your life?” I asked.

She thought for a moment.

“Was it when your grandpa died when you were four?”

“Yes.”

“Ok. That’s a 100. That was pretty awful, wasn’t it?”

“Yeah. I was really very sad.”

“I remember. So if that’s 100, where does your cow losing an eye go?”

She thought. “A seven.”

Then she shook her head. “No. A four.”

“Are you sure? You were pretty upset.”

“Compared to other things, Uncle Frank, it isn’t a big deal.”

“Exactly. That’s perspective.”

I told her to add a few more points to the scale: 80, 60, 40 and 20. Then to use it the next time she got upset, whether it was her brother, school or sports.

Anyone can use this.

100 is the worst event of your life. Not the worst thing you can imagine, but the worst event of your life. My 100 is the death of my grandmother when I was 19. I’ve met many people whose 100 is worse than mine. Everyone is different. We don’t compare traumas. If someone isn’t comfortable telling me or writing down their 100, I tell them to put down their 90 or 80. Then I ask them to figure out what a 60, 40 and 20 would be.

My 4Runner was stolen in Montreal in August of 2024 while I was on vacation with four friends. I had a lot of hiking gear in it as well. I was unhappy about it, but compared to my 100 and 90 and 80, it was about a 25. My friends asked what I was going to do. “We go on with our trip. We’ll go biking and then take a boat ride and get a great dinner. Just as planned. What else is there to do? Freak out? Ruin everyone else’s time?”

When I’ve taught it to my students or clients, I start off with the story of Chicken Little. It’s an old folk tale. The townspeople knew that the sky wasn’t falling; they all looked upon Chicken Little with great annoyance. There is a valuable lesson about human nature there: people don’t like being around people who catastrophize. It’s exhausting.

A student asked, “What if you have a client who says everything is 100?”

“Great question. For Chicken Little, everything was a 100. He had no perspective. Everything was a disaster. He ends up alone. This scale is most specifically for the people who rate every problem, every aggravation as 100. We use this to teach them perspective.”

Last week, a student asked me, “What if you haven’t been through much? That nothing bad had really happened in your life?”

“Well, good for you. Try to keep that up as long as possible. Don’t apologize for it though. Whatever your worst moment is,” I told him, “that is your 100. At some point, it will almost certainly be replaced.”

I have sat with many people during the worst moment of their lives. I always tell them that they are supposed to feel awful. Usually powerless, usually scared. There is no quick way out of it. For some of the worst moments, I tell them that they have the right to collapse, though I don’t recommend it. Instead, I recommend therapy, sleep, exercise, healthy eating, time with family and friends, enjoyable activities and just moving forward. Even if you don’t feel like it.

Sometimes, something that feels like a 75 today is a 30 a year later. “It wasn’t that big a deal,” or “I didn’t think I’d ever recover” and sometimes even “I learned a lot from that” are phrases I hear from clients.

When I look back on my grandmother’s death (100) or my friend Eric’s death (85) or my Dad’s (80), I can remember how terrible I felt. The sadness. The fatigue. The utter loss. But, I don’t feel that way now. They still are a 100, 85 and 80, but they don’t cause me serious distress. I don’t feel the way I felt when those events happened.

If someone keeps experiencing a 100 or a 90, months or years later, that is PTSD. You still feel like you are in the moment, experiencing the pain, long after the event has passed.

This isn’t treatment. It’s a way to keep everything from seeming like a 100.

I sent this article to my niece, who is almost 18 now. “I remember the lesson but I have almost no memory of the cow losing its eye. Which shows it really was a four on the scale and not important at all.”

AI: Thinking Less, Becoming Replaceable

I asked AI to write an article using a half dozen inputs. Then I wrote one using the same inputs. Both are below.

Economists have spent decades arguing that technology changes jobs more than it eliminates them. Automation replaces some tasks, creates others, and the workforce adjusts over time. That view is now shifting. Recent reporting shows a growing recognition that artificial intelligence is not just another incremental tool. It is affecting large portions of white-collar work in ways that are starting to resemble what happened to manufacturing.

In manufacturing, productivity gains reduced the number of workers needed to produce the same output. Fifty workers who produced a certain number of cars in 1970 can now produce far more. Demand did not increase at the same rate, so fewer workers were needed. Efficiency did not destroy the industry, but it reduced the number of people required to sustain it.

A similar dynamic is beginning to emerge in knowledge work. AI allows a single worker to complete tasks that previously required more time, more people, or both. Drafting, summarizing, analyzing, and communicating can all be done faster. On the surface, this looks like a clear benefit. Workers save time. Companies increase output. Stress decreases. Productivity rises.

But this is only part of the story.

Another recent analysis suggests that the vast majority of people are using AI in a very specific way: to reduce effort. They use it to write emails, draft documents, complete assignments, and move through their workday more quickly. A much smaller group uses AI differently. Instead of replacing effort, they use it to challenge their thinking, test assumptions, and improve their work.

These two approaches lead to very different outcomes.

When AI is used primarily to complete tasks, the worker becomes more efficient. But they also become more interchangeable. If the value of the job is defined by AI-assisted output, then multiple workers can produce similar results. Over time, this makes it easier for organizations to reduce headcount or replace individuals with others who can generate comparable work using the same tools.

In contrast, workers who use AI to improve their reasoning and judgment increase their value in a different way. They are not just producing output more quickly. They are producing better decisions, identifying errors, and adapting to new problems. Their work becomes less standardized and more difficult to replace.

This creates a divide within the same workforce.

On one side are workers who use AI to reduce friction. They complete tasks faster, lower their immediate stress, and meet expectations efficiently. In the short term, this improves their experience of work. In the longer term, however, it can flatten their skill set. If they rely on AI to generate answers without developing their own thinking, they risk becoming dependent on the tool in a way that limits their growth.

On the other side are workers who use AI to increase friction where it matters. They still benefit from speed, but they also use the technology to examine their own reasoning. They ask different questions. They compare outputs. They look for errors. Over time, this strengthens their ability to operate without the tool and to use it more effectively when needed.

The distinction is not about access to AI. Both groups have it. The difference is how it is used.

This is where the comparison to manufacturing becomes more precise. Efficiency gains reduce the number of people needed to perform a task. But in knowledge work, efficiency also changes the nature of the task itself. Work that can be standardized and accelerated becomes easier to consolidate. Work that depends on judgment, interpretation, and adaptation remains more resistant.

As AI continues to improve, these dynamics are likely to intensify. Tools will become more capable. Outputs will become more polished. The baseline level of performance will rise. At the same time, the gap between those who rely on AI for answers and those who use it to refine their thinking will widen.

Some workers will adapt. Others will not. Many will fall somewhere in between.

This is not simply a story about technology replacing jobs. It is also a story about how individuals respond to that technology. The same tool can lead to different outcomes depending on how it is used.

In the short term, using AI to complete tasks more quickly is appealing. It saves time. It reduces effort. It makes work more manageable. But over time, the habits that form around that use can shape what a worker is able to do without the tool.

Those habits matter.

AI will continue to change how work is done. It will increase productivity. It will alter expectations. It will reshape roles. The question is not whether these changes will occur, but how individuals position themselves within them.

I went to a speed dating event in Northern New Jersey in February with a friend. I instantly surveyed the landscape and recognized that I wasn’t interested in anyone, so I used the twelve women I talked to as a focus group to find out how they used AI.

A businesswoman who worked at Revlon told me that she used it to draft all of her emails at work, write reports and help make decisions. I asked her if she was concerned at all about losing skills or her boss finding out. She told me that she is more productive, less stressed and finally has time “to live my life again.”  I didn’t have the time or desire to tell her that if I were her boss, I would find a high school graduate who could use AI as well as her and pay them one-third her salary. I expect she’ll find that out herself in the next few years.

 A middle-school teacher from Montclair told me that she used it to write her lesson plans. “If it could grade papers and tests, I’d have it do that too,” she happily told me. I asked her if she felt less like a teacher because she wasn’t designing her course. She did not like that question at all. “I’m able to spend more time actually teaching the kids, so it makes me a better teacher,” she said, a bit defensively. Her job is probably safe for a while, because of a teacher shortage and the low expectations of the field. That safety removes the need to develop and improve. Hence the offloading of tasks that are key parts of teaching.  I suspect other teachers are doing this also, which means the level of instruction will slowly degrade over time. If true, this is dire for education.

A scientist from a German bio-tech company stated that she used AI to create, analyze and synthesize “just a massive amount” of excel spreadsheets. She said she uses three different AI systems and then cross checks their work to ensure there are no errors. “I am so much more efficient. The agents can create the data sets so much faster than any human worker. Taking away the grunt work allows me to utilize my other skills.” She was the only woman that night who I felt confident would still be employed in ten years.

All three of the women use AI to take care of tasks that they view as dull, tedious and time consuming. What I saw that night lines up almost exactly with what recent reporting is starting to show. A recent Business Insider article stated that at least 95% of workers use AI to think less, while the remaining 5% use it to think more. A New York Times article this morning quoted several economists who have finally come to the conclusion that AI probably will be very disruptive to the workforce, particularly if it continues to rapidly evolve and improve. The workers who are most in jeopardy of either having their salaries reduced or losing their jobs altogether are those who use it to think less. The businesswoman and the teacher are clear examples of professionals who have offloaded work to AI and are thinking less. The scientist uses AI as a collaborative partner; she uses it to challenge her assumptions and discuss the data before going to her team with it.

The professionals who are using AI to offload work, increase productivity and reduce stress are currently enjoying wonderful benefits from AI. But they are unknowingly removing the processes which once made them valuable. By using AI to increasingly complete their tasks, they are becoming more easily replaceable by another person who can do AI-assisted work. A professional with a deep knowledge base who uses AI to challenge their assumptions, test their models, edit their writing and expand upon their ideas is the rare worker who is likely to be inoculated from becoming redundant.

All three use AI. They all use it to complete some of their work. The businesswoman and the teacher are using it to think less. They are getting faster, but not better. The businesswoman will probably be the first to lose her job. The teacher’s job is safe, for now, but I believe that the quality of her work is already suffering. The scientist was the only one of the twelve women I met that night that uses AI to improve her thinking. The other 91% are actively participating in their own career extinction.

It took AI less than ten seconds to write a competent, passable article about how workers are using AI to complete tasks without increasing their knowledge or skills. Mine took about an hour.

AI explains the situation. It’s a boring, monotonous piece. It’s interchangeable. Any AI system could have written it.

Mine makes an argument. It uses real examples. It passes judgement. It’s entertaining and informative and could have only been written by me.

Some people would argue that both approaches work. The difference is obvious.

These are the choices workers are making every day.

Suicide: What Suicide Loss Survivors Face and What To Do

Michelle came to see me after her husband killed himself. He had retired a few years earlier after a long career with the NYPD. She was devastated. She couldn’t sleep. She was overwhelmed with grief, anger and guilt. She was a therapist. The questions “Why did he do it?” and “How could I have missed it?” raced through her head.

Put aside that he was a cop and she was a therapist. This is a common response to suicide. It’s a structure that traps people.

Suicide Loss Survivor is the term used to describe people who have lost someone to suicide. For many of us, it is like a nuclear bomb went off in our lives. Everything has been affected; everything hurts. Recovery often feels impossible. The Suicide Domino Theory states that survivors are at higher risk for suicidal thoughts and attempts than people who don’t know anyone that died by suicide.

Suicide produces questions without resolution.

  • Why did they do it?
  • What did I miss?
  • Could I have stopped it?

The mind attempts to treat these as solvable problems.

Edward Arlington Robinson published Richard Cory in 1897. No one in town can understand why this man, who seemingly had everything, killed himself.

Sophocles wrote about Ajax, one of the Greek heroes in the Trojan War, in the 440s BC. Ajax took his life after the death of Achilles, near the end of the war. His wife, his brother and his soldiers all struggled with his suicide.

Individuals and society have always struggled to understand suicide. These are not questions. They are unending loops.

The Structure of Suicide Loss

1. Complex Grief

  • Suicide is different from other deaths, because it often adds anger, guilt and confusion.
  • It’s non-linear. That means it doesn’t just heal with time.
  • grief + anger + guilt + confusion = complex grief

2. The Why Loop

3. The Self-Blame Statement

4. The Missed Signal Problem

  • Some people lament “I should have seen this.”
  • People try to reconstruct conversations
  • They look through old texts and emails
  • Clinicians frantically search through notes
  • They seek a clarity that cannot be found

5. Planet Pretend vs. Planet Suicide

  • Most people avoid talking about death, and they definitely avoid talking about suicide. They live on Planet Pretend.
  • Survivors are stuck in the complex grief. Their world has been smashed. Things are unresolved. They live on Planet Suicide.

The Suicide Loss Survivor Protocols

  1. Stop trying to answer why. It can’t be satisfactorily answered. It is not a question, but a doom spiral. People can learn to stop this by seeing a therapist or going to a suicide survivor support group. Once someone knows that this is a universal response, they can learn to recognize it and ultimately, stop asking it.
  2. Interrupt the Missed Signal Loop. Much like the just stated why question, survivors need to put a boundary on reviewing past interactions and replaying events.
  3. Structured Expression. I believe writing is the best way to deal with grief. These 30 prompts are a good start.
  4. Plan for dates and triggers. Birthdays, holidays and death anniversaries sometimes loom painfully large. It’s really important to come up with a plan for those days.
  5. Re-engagement. For many, laughing again feels like a betrayal to the dead. I work with people on re-engaging with life. I encourage them to spend time with family and friends. Go to religious services. Volunteer. Travel. Even date again.

Do Not

  1. Endlessly ruminate over the suicide and call it processing.
  2. Search for a single answer.
  3. Get validation from a therapist about your grief and never move forward.
  4. Avoid life.

I worked with Michelle for just about two years. She learned to stop asking why. She accepted that there was nothing wrong with her, that her husband’s suicide was not about her, even though it felt that way in the beginning. Michelle begrudgingly went out to dinner with other family members. Eventually she enjoyed those meals. She traveled with close friends. Near the end of our treatment, she went on a date. She felt guilty. She went anyway. It wasn’t a match. That was just fine though, because she had turned back to life.

Humans experience repeated loss. In Kurt Vonnegut’s Slaughterhouse Five, the phrase “so it goes” appears 106 times. It shows up after every death in the book. Vonnegut survived the trauma of the firebombing of Dresden. He wrote books that often used humor to handle heavy subjects. “I’d rather laugh than cry,” he once told an audience. Death is a constant part of human experience, and most of us will experience it many times. We cannot live in despair. We must move forward. Suicide resists the compression of “so it goes.”

We do not solve suicide. We live despite it.

AI and Clinical Practice: Spring 2026 Update

This is a summary of the AI and clinical landscape as of Spring 2026. The environment is changing quickly and standards are not yet settled. Regulatory and legal responses are uneven and still developing across jurisdictions. This update reflects ongoing clinical work alongside review of emerging law and literature. Its purpose is to reduce blind spots for clinicians and programs and to identify emerging risks to the profession and the public. Future updates will incorporate international developments as relevant.

State Regulation Is Moving First

Colorado HB26-1195. This is a bill that proposes to prohibit clinicians from using AI to directly communicate with patients. It also places restrictions on AI from creating treatment plans or therapeutic recommendations with a licensed professional reviewing them.

New York S7263. This bill imposes liability for damages caused by a chatbot imitating certain licensed professions, including counseling.

New Jersey S3668. This bill requires disclosure in certain AI-driven communication.

Washington State HB2225 and S5984. This bill mandates that AI chatbots inform users that they are not interacting with a human and then requires periodic reminders.

The Colorado bill is the most aggressive about protecting the public and clinicians while New York is laying the ground for user litigation against AI companies if they get unlicensed or bad advice. New Jersey and Washington are just about AI disclosure. There is no unified Federal standard. Clinicians are operating across inconsistent legal environments. None of them go far enough.

The Courts Are Beginning to Test Harm and Liability

Google was sued in Federal Court on March 4 by the family of a Florida man who took his life. The man’s family alleges that the Gemini AI chatbot encouraged his self harm.

The parents of a girl who was critically wounded in a mass shooting in Canada has sued OpenAI in British Columbia’s Supreme Court. They allege that OpenAI failed to act on warning signs from the shooters ChatGPT account.

    Liability theories are forming. Courts will shape foreseeability, duty to warn and liability. Regardless how these and other cases play out, they will influence practice, insurance and legal environments. I expect to see a significant increase in lawsuits, including families bringing claims related to harm, negligence and failure to intervene.

    Clinical Risk

    A new scientific review in the Lancet Psychiatry discussed how AI chatbots can encourage delusional thinking in vulnerable people. It describes how AI can present incorrect or reinforcing content. The biggest concern is that there is evidence that AI Chatbots can reinforce distorted thinking.

    How I Am Using AI

    I am using AI to edit some of my writing, generate counterfactuals and conduct hostile reviews of my articles. Any research that AI conducts I independently verify. When I do use AI research for an article, I cite it. I use AI to organize and analyze data and provide me with basic visualizations. I have used AI to synthesize themes and evaluate prior work in my career. I have a written doctrine about AI Use and Writing Standards available on my website that goes into much greater detail about this.

    I created a Student AI Use Guide for my Rutgers Seniors. At the end of the semester, I will post an updated version of that on my website.

    Provisional Practices

    Documentation and Writing: AI may assist with structure or editing. My best practice recommendation is that clinicians should not have AI write notes, treatment plans, treatment plan reviews or discharge summaries. Writing is both an act of thinking and learning. Clinicians who do all their own writing are much more likely to have a better understanding of their case, as well as keeping their skills sharp. The more one uses AI to write, the more one loses the ability to write.

    Diagnosis and Clinical Decision-Making: Do not rely on AI to determine diagnoses or treatment decisions. A medical student described using AI so frequently for diagnostic support that he realized he had gone an entire day without independently thinking through a single patient encounter (The New Yorker, “If A.I. Can Diagnose Patients, What Are Doctors For?”). He deliberately pulled back after recognizing the impact on his own thinking.

    Input Discipline: Do not enter identifiable client information into AI systems unless you are certain it complies with privacy laws and organizational policy. I believe there are likely widespread HIPAA violations happening all over the country in this regard.

    Supervision and Consultation: AI is not supervision. AI carries no ethical duty or liability. Even if the legal landscape changes, a good clinician should continue to seek human supervision and consultation. My abilities and career were shaped by my supervisors. Woe to the therapist who learns mostly or solely from a computer.

    Client Use of AI: Clinicians should ask their clients about AI use. They should find out if they are using AI for guidance or emotional support. College students have told me that they have friends who spend entire weekend evenings chatting with AI. One woman I talked to on a train told me that she was getting advice about how to handle her breakup from AI. To be clear, some people are using AI as a substitute for human interaction. Clinicians should evaluate the impact of a client’s AI use on symptoms, avoidance and functionality.

    Verification: Do not rely on unverified AI outputs. Any information used in clinical or professional work should be independently confirmed.

    AI Agents: Do not delegate communication or decisions to AI. This introduces legal and ethical risk.

    Disclosure and Risk Awareness: Assume AI-assisted content is discoverable. Check with your company’s policy on AI disclosure practices. That’s a bit of a joke, as most companies’ AI policies are outdated, incomplete or non-existent.

      In Closing

      AI is advancing faster than regulation, and only a few states are beginning to regulate AI as it pertains to the mental health field. Legal standards are going to emerge through court cases, both in the US and internationally. Many people are using AI for guidance and emotional support; there is a particular risk for people who are isolated or have delusional type thinking. Clinicians should be talking to their clients about their AI use and setting boundaries. AI can provide information and simulated interaction, but it does not replace lived experience, real-time clinical judgment or the relational work that happens between people. I wrote about AI’s potential impact on social work in 2024. At that time, I had not yet used these tools directly, but was already concerned about how they might affect clinical work and professional roles. The pace and scope of change since then has been significant. My plan is to provide updates on a quarterly basis.

      Bad Therapy

      For the third week of class this semester, my students read a story about a patient of mine that died. Bobby’s severe drug use had damaged his brain and he was painfully aware that he was diminished. At some point each session, Bobby would ask “When will my brain heal?”

      I always refused to lie. I told him that he may have done permanent damage but we would never really know until he had a few years of total abstinence from substances.

      That exchange shocked my students. They were uncomfortable with the concept of delivering bad news. “You didn’t soften it,” one said.

      “That would be bad therapy,” I responded.

      Another student softly muttered, “I need to get a new therapist.” A few others nodded. Many of my students get bad therapy. Between their personal experiences and, at times, woeful supervision, they are being trained this way.

      Bad Therapy often looks like:

      • Premature reassurance
      • Incorrect validation
      • Ideological Collusion
      • Passing over key moments
      • Avoiding core questions
      • Avoidance due to clinician discomfort
      • Diagnostic inflation
      • Failure to ask foundational questions
      • Failure to review records
      • Failure to reassess diagnosis
      • Failure to recognize origin patterns
      • Forced insight and change too quickly

      Some people are in therapy for three years and have not moved an inch.

      A 28-year-old woman was anxious about a job she applied for. Her therapist responded, “You’ll get the job. They’d be crazy not to hire you.” The therapist promised an outcome that was actually uncertain. A temporary sense of relief. This is premature reassurance.

      “If you get the job, good for you. If you don’t it, we’ll discuss your next moves and apply for some other jobs.”

      A 33-year-old woman often complained in therapy about her mom, boss, co-workers, friends and last three boyfriends. The therapist agrees with each complaint. This is incorrect validation.

      The therapist never names the one thing they all have in common.

      A right-wing client goes to see a right-wing therapist. The client complains about liberals, immigration, crimes and pronouns. The therapist agrees that America is going down the toilet.

      A progressive client goes to see a progressive therapist. The client complains about MAGA, racism and the cost of housing. The therapist agrees that America may no longer be a Democracy. This is ideological collusion, where therapy becomes an echo chamber. Agreement replaces agency.

      A group of high school athletes meet with a school counselor to talk about managing their time. One mentions that he is “afraid of my Dad.” The clinician doesn’t stop to address it. This is passing over a key moment.

      A couple goes to couples therapy every week for nine months. They argue both in and out of session. At no point did the clinician ask, “Do you want to stay married?” The man didn’t. Nine months had been wasted. This is avoiding core questions.

      Well over a decade ago, I listened as a psychologist did a case presentation on a 26-year-old lesbian who had recently left the Navy and was new to Rutgers. She was having relationship problems. I asked her how often the client and her partner were having sex. “I didn’t ask about that,” the psychologist answered. I told her that it was an important question. This is avoidance due to clinician discomfort.

      A 23-year-old man goes to a therapist to deal with the death of his older brother several years earlier. The therapist never asks, “How did you find out?” This is one of the most important questions when dealing with grief early on, as finding the body is different than getting a text message which is different than someone calling you on the phone and telling you directly. This is entry trauma ignored, which is often avoidance due to clinician discomfort.

      Diagnostic Errors

      An 18-year-old female walked into my office at Rutgers in the fall of 2009. “I’m bi-polar and have borderline personality disorder. I have anxiety, depression, ADHD and PTSD. I am in recovery from alcohol, marijuana and hallucinogens.”

      “Let’s back up a minute. Tell me your name and where you are from,” I countered.

      She had walked in with seven diagnoses. Her identity was wrapped around pathology. She later told me she felt irrevocably broken. This is diagnostic inflation.

      I was hired by a program in New York City to help about a half dozen clinicians improve their skills and work on their case presentations. After the first one presented a case involving insomnia, I asked, “How long has this been going on?” He couldn’t answer. Another clinician presented on about an abusive relationship. “What do her friends think?” She couldn’t answer. A third clinician discussed a client’s overly anxious responses. “Where did they grow up? Are her parents alive?” They couldn’t answer. These were blatant failures to ask foundational questions.

      A parent of a 30-year-old man at a treatment program that I was consulting with in Pennsylvania contacted me with concerns about her son. Based on her description, it was clear that he had schizophrenia. Years of records were available, but the treating clinician had never obtained them. They only had obtained the records from the referring detox. Even in other areas of medicine, there is sometimes a failure to review prior records.

      When I was in the Army, a 25-year-old female soldier mistakenly walked into my office. She was being medically discharged from the Army because she had been diagnosed with bi-polar disorder. She had caught her boyfriend in their bed when she came home a day early from training. She broke up with him, packed up her stuff and moved back in with her parents. She had a hard time sleeping for a few days and couldn’t get her mind to stop racing. She went to the emergency room at Hershey Hospital and the doctor diagnosed her with bi-polar disorder. Ridiculous. She wasn’t bi-polar, she was upset from a break up. I reassessed her and she was able to stay in the Army. Therapists take in clients and often just reaffirm diagnoses instead of reassess them. Diagnosis is not a permanent label, but rather an ongoing process. All too often, therapists and other medical professionals fail to reassess diagnoses.

      Just before COVID, a 43-year-old cop came to see me. He was miserable at work; so much that his wife wanted him to quit after 18 1/2 years. He only needed to work another one and a half years to get his pension. He had seen a few other therapists and nothing had helped.

      “Well,” I told him, “we aren’t quitting this close to retirement.” I went through his background and learned his father was an abusive alcoholic. I told him that I wanted him to go to three Adult Children of Alcoholics (ACOA) meetings in the next month. When he returned, he had gone to 20.

      “After the fourth one, I called my younger sister and said, ‘it wasn’t us.'” Dealing with his family of origin issues lowered his stress baseline and he was able to survive his last few years at work. He actually lasted another four years before he retired. This was an example of failure to recognize origin patterns.

      In 2010, a 25-year-old male graduate student came to me for help with his anxiety and “bitchy” girlfriend. I recognized his drinking problem within 15 minutes. I told him that I wanted him to attend Early Recovery Group on Tuesdays and check out a few on-campus AA meetings with some other students. He disappeared for six weeks.

      I told my supervisor what happened.

      “He left because you were forcing him to a place he didn’t want to go. He didn’t see alcohol as problem. He wanted to work on his anxiety and relationship.”

      “Those aren’t his primary issues,” I responded.

      She calmly smiled. “To him, they are.”

      He returned after six weeks and we worked on his anxiety and romantic relationship fights. After 18 weeks, he started to see the connection between those problems and his drinking. My mistake was an example of forced insight and change too quickly.

      I have worked with people who took six months to realize what I knew in 15 minutes. That used to feel like wasted time. Now I accept that it sometimes takes people a long time to recognize their problems.

      Some people sit in therapy and never move. A bad therapist won’t push or challenge them, point out when their behaviors aren’t matching up with their values and often engages in going-nowhere-validation therapy.

      The Greenagel Stress Volcano

      From the Greenagel Equations

      The Greenagel Equations are a set of practical frameworks developed between 2005 and 2008 in schools, outpatient and family treatment settings. They were built in rooms, not in theory, and have been used with students, families, law enforcement, veterans and therapists.

      Everyone has a baseline level of stress. Your baseline depends on a few factors: childhood experiences, parental modeling, adult experiences and your stress regulating behaviors. This model explains how stress builds and when it turns into a crisis.

      Stress functions like a volcano. Ideally, the magma level is one’s baseline stress. As we experience stress, the pressure rises. When we experience a crisis, it is like a volcanic explosion. Ash darkens the sky and lava runs down the sides, threatening all in its path.

      Some people live near the bottom. Some live near the top. Most live somewhere in the middle. The lower one’s baseline, the greater their capacity to handle stress and emergencies without blowing up. Higher baselines mean that small triggers can lead to an explosion.

      As a teenager, I lived near the top and would easily erupt over petty and inconsequential things. Entering recovery at 19 and finding a variety of calming behaviors moved my baseline to the low magma level. This has helped me survive my work.

      Everyone experiences increases in stress. Movement up the volcano is normal, for both small and big things. During big Vikings or Dodgers games, my stress level goes up a little bit. The key here is baseline position, not the existence of stress.

      Unlike the other equations, I don’t remember the exact moment I came up with the volcano. I know it happened in a substance misuse treatment group though. Clients would talk about how they would experience stress and act out sexually or binge eat or overspend. And while it helped them temporarily, eventually the good feeling passed and their stress level would rise again, sometimes even worse than before because of their actions. They would move down the volcano when they were high, but the next day they would rise right back up to their baseline level. Eventually, I included all of the process disorders as unhealthy ways that people try to manage their stress:

      • Substances
      • Gambling
      • Video Games
      • Screens & Social Media
      • Food
      • Shopping
      • Sex & Pornography

      Two consistent factors come into play here:

      1. The relief from these behaviors is temporary
      2. Engaging in these behaviors over a long period of time leads to an increase in the baseline.

      Whenever I teach this, whether it is in group, class, trainings or individual sessions, I wait for someone to ask, “How do I permanently move down the volcano?”

      I always smile and say, “I was waiting for that question.”

      I tell people that these behaviors will help people slowly move down the volcano. They need to be consistently done over a long period of time:

      • Therapy
      • Reducing or, ideally stopping, maladaptive behaviors (drinking, gambling, overeating)
      • Exercise
      • Proper sleep hygiene
      • Healthy eating
      • Time off from work or school
      • Time in nature
      • Avoiding red arrow people
      • Participation in sports, clubs, associations and 12-step groups
      • Enjoyable hobbies
      • A sense of purpose

      Consistent engagement in a number of these will move someone down the stress volcano.

      For people who had a rough childhood, their baseline stress as an adult will be higher. If you have been trained on or read about ACE scores, you know that a higher ACE score predicts a vast number of physical and psychological problems. That written, ACE is not destiny. Someone can have a high ACE score and still end up being pretty low on the stress volcano. But, I can assure you, they engage in several of the aforementioned positive behaviors.

      For people in high stress jobs, their baseline stress will be higher. High stress jobs include but are not limited to the military, law enforcement, fire fighters, EMTs, and emergency room medical professionals. To be clear, not all military and law enforcement jobs are stressful. The highest type of stress comes from a) your own life being in danger b) protecting/rescuing other people in danger c) seeing awful things happen to kids and d) being continually responsible for a number of people.

      If you are reading this and identify that you had a rough childhood and you are in a high stress job, you should be on high alert that you are at risk for living at the top of the stress volcano. You might have figured this out already though and have engaged in a bunch of those positive behaviors for years as a way of taking care of yourself.

      A couple of years ago, I taught my stress volcano at a State Police Forensic Lab and I later learned that a number of the attendees weaponized it. “Your stress volcano is super high” and “you are making me move up the stress volcano” were phrases employees threw at each other.

      I don’t want people doing that. This is not for you to label others or criticize them. It is designed for self-assessment.

      Now, if you are in the early stages of a relationship and recognize that your romantic partner lives at the top of the volcano and you are tired of dealing with the explosions, then by all means, leave.

      Where is your baseline?

      What raises it?

      What lowers it?

      The goal is not to eliminate stress. The goal is to lower where you live.

      The Substance Misuse Spectrum

      From the Greenagel Equations

      The Greenagel Equations are a set of practical frameworks developed between 2005 and 2008 in schools, outpatient and family treatment settings. They were built in rooms, not in theory, and have been used with students, families, law enforcement, veterans and therapists.

      Substance misuse exists on a spectrum from non-use to severe disorder, defined by behavioral, physiological and consequence-based criteria.

      I was running a group at Integrity House in 2004 when a client exploded in anger, “I’m tired of all these people telling me I have a problem. You an addict. You an alcoholic. You need treatment. And they keep sending me away to these bitch ass programs.”

      “Has anyone ever explained to you how we diagnose someone with substance abuse or dependence?” I asked him.

      He immediately calmed down. “No.”

      I listed out the criteria from the DSM IV-TR on the board and broke down what each one meant.

      Group members started saying what they did and didn’t have. Some confronted each other on their denial. It was a wildly effective group.

      I didn’t like how substance abuse had four criteria and substance dependence had seven with only role failure in both. I explained that role failure meant that substance use caused a problem at work, school, with family, with friends, with a romantic partner, legal issues, financial difficulties or health complications.

      “All of those?” someone asked.

      “No, just one life area needs to be affected.”

      I created a spectrum, starting with non-use on the left and then use, abuse and dependence as it moves right. I used role failure as the anchor. Both people who used and those who didn’t use substances were united that they didn’t have any role failure caused by using alcohol or drugs. They may have problems, but they weren’t caused by substances.

      For the nine years, I taught substance misuse as a spectrum, rather than two separate diagnoses. In 2013, DSM-5 came out. It threw out the terms abuse and dependence, got rid of legal problems for diagnosing and listed a clean set of 11 criteria. The APA had come around to seeing this as a continuum, or spectrum.

      I wanted to include percentages for each group. I looked up data and crosschecked it. For years, it was roughly 20% of adult Americans who do not use alcohol or drugs at all. That about 50-60% use without having any problems and 20-30% meet criteria for a substance misuse disorder. In the 2020s, various data sets place those numbers at 20-30%, 50-60%, and 15-20% respectively.

      I have trained and assisted staff and students on the RU Sure Campaign for just about two decades. Their research at Rutgers states that

      2 out of 3 students drink 3 drinks or less

      and that

      1 in 5 don’t drink at all

      This breaks down to 20% non-users, which is consistent with my aforementioned data. Whenever I have presented these numbers, especially to people in treatment, correction settings, lawyers, cops or military, I am inevitably met with some version of this comment: “Yo man, your numbers are bullshit.”

      People who drink heavily often cannot conceive of that 20% of adults don’t drink at all. “You ever hear the phrase birds of a feather flock together” or “You are identified by the company you keep?”

      People nod.

      “When I was 15, I was on juvenile probation. My father was horrified. I told him that over half the kids I knew were on probation.”

      He responded “juvenile delinquents hang out with other juvenile delinquents. Behavior doesn’t happen in a vacuum.”

      Two rules to consider with the spectrum:

      1. If I take 1000 non-users and 1000 users and then look at them in ten years, there will be more people that meet criteria for a substance misuse diagnosis from the using group than the non-using group. This holds true throughout the spectrum. So, the further along you are, the more likely you are to continue.
      2. One of the first people I ever treated was a 53-year-old white female who was a middle school music teacher. She met three of the four criteria for substance abuse. She was unhappy, but didn’t have any work or legal or health problems from her drinking. About a month later, I was treating a 15-year-old boy who also met three of the four criteria. He was having problems in school, had been arrested and his parents were both furious and worried about him. His life was already upended and it was very likely that it was going to get worse if he didn’t stop. So, if someone is young and already meets diagnostic criteria, it is much more likely to get worse than someone who is older.

      This is the single lesson I have taught more than anything else in my life. I have taught this in high schools, universities, jails, prisons, hospitals, in-patient and out-patient settings, community centers, Army bases, with individual clients and a number of other settings.

      Substance misuse exists on a spectrum from non-use to severe disorder, defined by behavioral, physiological and consequence-based criteria.

      This is how I explain the criteria in group settings.

      Role failure – One challenging aspect is how people define failure. Some students are good with a C, while another student may view a B+ as a total failure. I think it is a good idea for someone to come to a consensus with a counselor on a fair definition of role failure.

      Use despite problems – Healthy people cut down or stop when their substance misuse has caused a health, legal, financial or any other type of problem. A doctor tells you to stop drinking and you counter with reducing your drinking for the time being. A counselor tells you that the anti-anxiety medication will be less effective if you use marijuana and you smoke anyway.

      Tolerance – It takes more of the substance to get the same effect, or you experience a weaker sensation with the same amount. People will often use a harder version of the substance (80 proof alcohol instead of beer or 74% THC content in a vape instead of 15% THC in the marijuana plant) or combine substances.

      Withdrawal – Most people think of shakes and hallucinations when they think of alcohol withdrawal. Those are late stage alcoholism withdrawal features though. More common are headaches, gastrointestinal issues, fatigue and irritability. The most common physical withdrawal feature of marijuana is sleep problems. Psychological withdrawal includes difficulty concentrating, confusion, memory issues, irritability and lack of motivation.

      Confrontation (x2) – The DSM-5-TR asks if you have one person in your life who thinks you have a problem. I don’t like that, as there may be someone who has recently quit drinking and wants to change everyone else or is some kind of evangelical teetotaler. I don’t consider unified parents to be two people; I’m also looking for a teacher or coach or friend or romantic partner or the court system to state their concern.

      Use of motor vehicle or machinery or other dangerous activities – The number one instinct in all creatures is survival. If you operate a motor vehicle or other machinery while under the influence, you put yourself at increased risk of injury or dying. This goes against nature. Or you get into more dangerous situations than you would when you were not using, you meet the criteria.

      Reducing/changing friends and activities – If you used to play softball, sing, paint or ride horses and now you do those things a lot less or not at all, we need to examine whether your use led to this. If you gave up friends that didn’t drink like you or questioned your marijuana use, you probably meet the criteria.

      More use, more time than planned – If you bought a 1.75 liter of vodka and planned it to last all of spring break but it ran out on Wednesday, you’re hit. If you acquired 90 Adderall pills for the semester and are out of them by mid-terms, you meet the criteria. If you planned on smoking until 10 pm but you didn’t stop until after midnight, you got it.

      Previous attempts to cut down or quit – You once tried to give up marijuana for Lent and you got to day 15 and then said, “Jesus will understand.” You stopped drinking hard liquor and switched to beer, only to eventually get back on hard liquor or drinking an excessive amount of beer. You said only on the weekends and the weekends started on Thursdays. No more bongs, just blunts. This is my last vape. Any of these and you meet the criteria.

      30 Hours or more in a week – A week is made up of 168 hours. If you have ever had more than one week (I’ll give you Jamaica 2014) where you were under the influence for 30 hours, you have it. Family members and friends are often aghast at the 30 hours number: “It’s like a full-time job.” Clients I’ve had at group often scoff at the same figure: “I get to 30 hours by Tuesday.”

      Physical or psychological problems caused or exacerbated by use – If you got drunk and sprained your ankle. If you had bronchitis and smoked anyway and it lasted for two months. If you never were depressed but now you are down unless you are using, or you had anxiety and you find that since you started smoking, your anxiety has gotten worse.

      How to Handle Death Anniversaries and Holidays: Make a Plan

      I recently recognized that I haven’t written about how to handle death anniversaries and holidays in any of my grief articles, despite the fact that this is something I work on with all my grieving patients.

      The first year after a death is a massive readjustment period. The first Father’s or Mother’s Day, Thanksgiving, Christmas, Hanukkah, Eid al-Fitr and birthdays often loom large. Common thoughts and statements include:

      • “How do I celebrate now?”
      • “I miss him.”
      • “It won’t be the same.”
      • “I don’t want to do anything.”
      • “Thinking about it makes me upset.”

      You have the right not to celebrate. You have the right to not plan anything. You have the right to sit in grief.

      However, I think those are very ill-conceived choices.

      Without addressing grief, without addressing looming dates, people often not only struggle with holidays, but regular numerical reminders and whole seasons. Unaddressed grief intensifies and spreads.

      One client’s sister took her life on March 13th. His father posts pictures and stories on Facebook on the 13th of every month. Another client’s husband took his life in October, and for almost two years, the entire Fall season was difficult.

      Besides individual counseling, grief groups, writing, connecting with others and sometimes finding new purposes, I work with clients on planning for upcoming holidays and anniversaries. Here is the playbook:

      1. Make a plan to do something on the holiday or birthday or first death anniversary.
      2. Negotiate and discuss with others what that plan will be. Even if there are other grieving family members who don’t want to do anything or are indecisive, you should still make a plan.
      3. Whatever the plan is, do it in the morning. Otherwise, much of the day is spent thinking and worrying over how it is going to go, how you will feel and how others might respond. If there is a big family meal in the late afternoon or dinner, I think it is wise to say something about the departed, that you miss them and love them and, hey, remember when they said or did this at the last holiday? Give people space to laugh or cry or sit uncomfortably or run out of the room. Do not lengthen or manage it. Move on with the holiday.
      4. For the first death anniversary: installing a tombstone, visiting a grave and reading or telling stories are all excellent plans. After the first death anniversary, I don’t suggest continuing to mark the date. However, if one can find a purpose with the date, that might work. For a few years, I had my friend Eric Arauz come and talk to my high school students about addiction on my friend Frazier’s death anniversary.
      5. In January, my Mom texted me that it was her Mom’s death anniversary and that she was sad. She died in 1979. I told her that it was probably more about her brother Bryan, who died in March of the previous year, making this the first time that she didn’t have a sibling to talk to about missing her parents. I suggested that she engage in some activity her Mom liked and go do something positive. She recognized that she was grieving her brother. This is common; anniversaries often activate other losses. She ended up going to the gym and went swimming.
      6. I’ve told people to eat some food that their Dad liked on his birthday (as long as they liked it too). I’ve had others do things with their daughters that they used to do with their sisters. This shifts grief from passive recall to active behavior.
      7. If you feel overwhelmed, you can have a long cry. But, do not stay in bed all day. Do not isolate. Get up, shower and go be around other people. Even if you don’t want to. Go do something even mildly productive.

      I’m currently running a grief group. During the first session, many of the participants stated that they were comforted that they weren’t alone. “I have been so overwhelmed with sadness and I’ve also been angry at the rest of the world for just moving along,” one member started, “and to hear that other people experience this makes me feel a little bit better.”

      Early on, grief can feel insurmountable. Looking back on it, we can remember it was painful but perhaps not even grasp how debilitating it felt. Grief writing plus therapy plus planning can help process and reduce the intensity over time.

      Do not allow birthdays, holidays and death anniversaries to dominate your thinking or fill you with dread. Make a plan and follow through.