Open Dialogue, Jaakko Seikkula, Tom Andersen, Mary Olson, ISPS, and the Cloud AI Scribe Vendor Archive: Credential Landscape, Vendor Archive Record Types, and the HIPAA Privilege Gap

Credential landscape: Jaakko Seikkula, Tom Andersen, Mary Olson, the Institute for Dialogic Practice, ISPS, and the non-licensed Open Dialogue practitioner population

Open Dialogue has no dedicated governmental credentialing body in the United States. No US state has enacted an Open Dialogue-specific licensure statute. No governmental body maintains a mandatory registry of Open Dialogue practitioners with enforcement authority over their clinical activities or their session documentation. Practitioners delivering Open Dialogue in licensed clinical settings hold state-issued clinical mental health licenses — Licensed Clinical Social Worker (LCSW), Licensed Professional Counselor (LPC), Licensed Marriage and Family Therapist (LMFT), or licensed clinical psychologist — or, in the case of psychiatric nurses, community mental health workers, and peer support specialists who participate in Open Dialogue teams, they may lack qualifying clinical mental health licensure entirely, in which case their session records carry no psychotherapist-patient privilege protection.

Jaakko Seikkula, PhD, is the primary developer of Open Dialogue. Seikkula is Professor of Psychotherapy at the University of Jyväskylä, Finland, and has held academic appointments at the University of Tromsø, Norway, and the University of Agder, Norway. Seikkula developed Open Dialogue at Keropudas Hospital in Tornio, Western Lapland, Finland, in the late 1980s and through the 1990s, building on Yrjö Alanen’s Need-Adapted Treatment model and on the dialogical philosophy of Mikhail Bakhtin. Seikkula’s foundational publications include the five-year outcome study of first-episode non-affective psychosis (Seikkula, Aaltonen, Alakare, Haarakangas, Keränen, and Lehtinen, 2006, Psychotherapy Research, 16(2), 214–228) and, co-authored with Tom Erik Arnkil, Dialogical Meetings in Social Networks (Karnac Books, 2006). Seikkula’s academic affiliations are European universities with no HIPAA authority and no US governmental health oversight status under any provision of 45 CFR Part 164.

Tom Andersen, MD (1936–2007), was Professor of Social Psychiatry at the University of Tromsø, Norway. Andersen developed the reflecting team technique — in which the treating clinical team conducts a narrated reflection dialogue about the patient and family in the patient’s and family’s presence — and published the foundational paper in 1987 (Andersen, T., 1987, “The Reflecting Team: Dialogue and Meta-Dialogue in Clinical Work,” Family Process, 26(4), 415–428) and the primary text The Reflecting Team: Dialogues and Dialogues About the Dialogues (Norton, 1991). The reflecting team technique became a core structural element of Open Dialogue’s network meetings. The University of Tromsø is a Norwegian public research university with no HIPAA § 164.512(d) authority over practitioners in the United States.

Birgitta Alakare, MD, psychiatrist at Keropudas Hospital and later at the Western Lapland Hospital District, was a key clinical co-developer of Open Dialogue at its founding site, contributing to the outcome research and clinical protocol development. Her institutional affiliation is a Finnish public health authority with no HIPAA § 164.512(d) authority in the United States.

In the United States, Mary Olson, PhD (Smith College, Northampton, Massachusetts), played a foundational role in introducing and developing Open Dialogue. Olson is a founding member of the Institute for Dialogic Practice, a private US training organization based in Northampton, Massachusetts, that offers training in Open Dialogue and dialogical practices for US clinicians. The Institute for Dialogic Practice is a private training organization — not a US governmental entity of any kind, not a US governmental health oversight agency, and not an entity with HIPAA § 164.512(d) authority to compel disclosure of practitioner session records from commercial cloud AI vendors. Open Dialogue – US (OD-US) is the North American Open Dialogue training and dissemination initiative; it is a private training program, not a governmental regulatory body.

The International Society for Psychological and Social Approaches to Psychosis (ISPS) is a private international nonprofit professional membership organization founded in 1956 and incorporated as a UK charity, with national chapters in numerous countries including ISPS-US. ISPS promotes psychological, social, and family-based approaches to psychosis and related conditions. It is not a US governmental entity of any kind. It is not a US governmental health oversight agency. It has no HIPAA § 164.512(d) authority to compel disclosure of practitioner session records from commercial cloud AI vendors. HIPAA § 164.512(d) applies to US covered entities making disclosures to US health oversight agencies conducting authorized oversight activities of US covered entities — a category that does not include a private international professional membership society promoting psychological approaches to psychosis. ISPS issues no clinical credentials, no board certification, and no licensure. ISPS membership does not confer psychotherapist-patient privilege on any member’s session records.

The practitioner population delivering Open Dialogue in the United States includes licensed clinical social workers (LCSW), licensed professional counselors (LPC), licensed marriage and family therapists (LMFT), licensed clinical psychologists, and psychiatrists — and also a substantial population without qualifying clinical mental health licensure: psychiatric nurses (registered nurses (RN) who participate in crisis teams and community mental health teams — RN licensure is a nursing license, not a clinical mental health license, and does not confer psychotherapist-patient privilege on session records in any US state); community mental health workers and case managers without qualifying clinical mental health licensure; peer support specialists (state-certified peer support workers have no qualifying clinical mental health licensure in any US state and their session records carry no psychotherapist-patient privilege); and pre-licensed trainees in supervised training programs who participate in Open Dialogue networks under supervision. Because Open Dialogue is explicitly a team-based approach in which the treatment team frequently includes two or more practitioners simultaneously — and the team composition varies across network meetings — a single series of vendor archive network meeting records for a named client may be generated by practitioners spanning licensed and non-licensed members of the same treatment team, creating a vendor archive record set with mixed privilege status in a single treatment course.

Treatment development: Open Dialogue’s origins in Western Lapland Finland, the seven principles, the reflecting team, dialogism, and outcome evidence

Open Dialogue emerged from the clinical research program at Keropudas Hospital in Tornio, Western Lapland, Finland, in the late 1980s and 1990s. The Western Lapland region developed what became known as the Western Lapland Treatment Model for first-episode psychosis — a community-based approach built around immediate crisis response, intensive social network inclusion, and the avoidance of premature diagnostic closure and hospitalization. The model drew on two primary intellectual lineages: Yrjö Alanen’s Need-Adapted Treatment approach (Alanen, 1997, Schizophrenia: Its Origins and Need-Adapted Treatment, Karnac Books), which emphasized individualized treatment planning tailored to the specific needs of the patient and their network rather than uniform protocol-driven care; and the dialogical philosophy of Mikhail Bakhtin (particularly the concept of the polyphonic novel, in which no single voice is authoritative and meaning emerges through the interplay of multiple voices), which Seikkula and colleagues applied to the therapeutic encounter as a framework for understanding why multi-voice network meetings in which no voice — not even the treating team’s — is privileged as authoritative might produce better clinical outcomes than expert-driven individual treatment.

Open Dialogue is organized around seven core principles:

Immediate help: The first meeting is organized within 24 hours of the initial crisis contact, at whatever location is most appropriate — the named client’s home, the hospital unit, the community center — on the same day if possible. The person who receives the crisis call takes responsibility for organizing the first meeting.

Social network perspective: The named client’s social network is included from the first meeting. The aim is to identify and invite every person who is significant to the named client and to the named crisis situation — family members, friends, neighbors, employers, teachers, social workers, housing workers, and any other named persons whose perspective on the named client’s situation is relevant. The social network is not treated as a backdrop to the individual’s treatment; it is treated as the therapeutic context in which recovery happens.

Flexibility and mobility: The treatment team goes to wherever the named client and named network are — home visits, community settings, hospital rooms, wherever the named crisis is taking place. The named client’s and named network’s needs determine the format, frequency, and location of each meeting.

Responsibility: The Open Dialogue team takes full responsibility for the treatment coordination from the first contact — no referral to a different team, no handoff of the named client to a specialist. The same team maintains psychological continuity across the treatment course, which may span months or years.

Psychological continuity: The same treatment team remains with the named client and named network across the treatment course, maintaining continuity of relationship and of the developing dialogical understanding of the named client’s situation.

Tolerating uncertainty: The Open Dialogue team explicitly avoids premature diagnostic closure. The model treats the diagnostic process as a potential obstacle to dialogue if it forecloses the multiple voices and multiple understandings present in the named network. The team’s documented stance is to hold multiple named explanations for the named client’s crisis in suspension — to tolerate not-knowing — rather than moving quickly toward diagnostic labels that may shut down the network’s own meaning-making process.

Dialogism: The primary therapeutic mechanism is the generation and sustaining of dialogue — multi-voice, polyphonic conversation in which no voice is authoritative, no conclusion is premature, and the emergence of new shared language is the measure of clinical progress. This is the Bakhtinian core of Open Dialogue: the therapeutic encounter as a space where multiple voices can speak, be heard, and find new language for the named crisis situation.

The reflecting team technique, developed by Tom Andersen and incorporated into Open Dialogue’s standard network meeting format, divides the meeting into structured segments in which the named client and named network speak while the clinical team listens, and then — with the named client and named network watching and listening — the clinical team speaks among themselves in an audible, narrated reflection on what they heard and what it evoked for them, before the named client and named network are invited to respond. The reflecting team’s reflections are offered tentatively, in the first person, as associations and hypotheses rather than as expert interpretations or diagnostic conclusions. The named client and named network then respond to the reflecting team’s narrated reflections, and further dialogue unfolds. This structured alternation between the named network’s voice and the named team’s reflecting voice creates the polyphonic session format that is Open Dialogue’s structural signature.

Open Dialogue’s outcome data from the Western Lapland program are among the most striking in the psychosis treatment literature. The five-year outcome study (Seikkula et al., 2006) of first-episode non-affective psychosis treated with Open Dialogue reported that approximately 75% of participants had no residual psychotic symptoms at five-year follow-up, approximately 80% were employed or studying, and antipsychotic medication had been used in approximately 35% of cases with the remainder treated without antipsychotics throughout the treatment course. These outcomes substantially exceed international standards for first-episode psychosis outcomes in standard treatment, though the Western Lapland program’s specific regional context has been noted as a potentially important factor in replication efforts. Randomized controlled trial research on Open Dialogue outside Finland is ongoing, including the ODDESSI trial in the UK (National Health Service, beginning 2017) and the ODTCC trial in Canada. The outcome evidence, while striking, does not affect the privilege analysis — the structural novelty of Open Dialogue’s vendor archive records exists regardless of the approach’s efficacy.

Vendor archive record type 1: Open Dialogue network meeting session narration

The Open Dialogue network meeting session narration is the vendor archive session record generated when an Open Dialogue practitioner or team uses a cloud AI scribe to document a network meeting — the clinical encounter in which the named client’s entire named social network is invited to be physically present simultaneously, including named family members, named friends, named neighbors, named employers, named colleagues, named housing caseworkers, named teachers, named school counselors, and named other support persons — and each named attendee’s in-session verbal contributions are documented as primary clinical content. It is the only vendor archive session record in 281 posts organized around a clinical encounter in which the full named social network is present and each named network member’s verbal characterizations of the named client’s situation, named relationships, named behaviors, named history, and named current crisis circumstances are primary clinical content in the same session record.

The network meeting session narration documents: the named meeting attendees (each named by name and named relationship to the named client — named mother, named employer, named neighbor, named housing caseworker, named former partner, named teacher, named peer support friend); each named attendee’s verbal contributions to the named dialogue (what each named person said about the named client’s named behaviors, named relational patterns, named living circumstances, and named crisis presentation); the named client’s responses to each named network member’s contributions; the named team’s named dialogical facilitation moves; and the named location where the meeting occurred (the named client’s home address, the named hospital unit, or another named setting). Each of these elements names persons beyond the identified named client — named employers, named neighbors, named caseworkers — and documents what those named third parties said about the named client in a clinical context, in a commercial cloud AI vendor’s archive, accessible through third-party subpoena when the named practitioner lacks qualifying clinical mental health licensure.

This is structurally distinct from all 280 prior posts in this series in a fundamental way. All 280 prior posts document sessions with a named identified client and optionally a named intimate partner or named family members in couple or family formats. The most expansive prior format in the series is the multi-family group sessions covered in DBT-A (post #218, which documents named adolescent and named parent in a multi-family skills training group) and the family therapy formats covered in SFT (post #273), BSFT (post #233), ABFT (post #227), FFT (post #229), MST (post #230), MDFT (post #231), CPP (post #228), and the Milan System (post #279). None of these formats generates a session record that simultaneously names the identified client’s named employer, named employer’s characterization of the named client’s work performance, named neighbor’s characterization of the named client’s behavior, named housing caseworker’s characterization of the named client’s living circumstances, named school counselor’s characterization of the named client’s academic functioning, named former partner’s characterization of the named client’s relational history, and named family members’ characterizations of the named client’s named childhood history — all in a single session record from a single clinical encounter. The Open Dialogue network meeting session narration is the only record type in 281 posts that does this.

The network meeting session narration’s scope of third-party naming has specific significance in the vendor archive. A single Open Dialogue network meeting narration may name ten to twenty people beyond the identified named client, document what each named person said about the named client’s named psychiatric crisis, and document those statements in a commercial cloud AI vendor’s archive accessible through third-party subpoena by any party in any legal proceeding in which those statements are material. The named employer who attended a network meeting and described the named client’s declining work performance in the weeks before the crisis; the named neighbor who described the named client’s behavior in the shared building; the named housing caseworker who described the named client’s recent housing instability; the named teacher who described the named client’s recent school absences — all of these named persons’ named statements appear in a single vendor archive session record that is not covered by psychotherapist-patient privilege when the named practitioner or named team members lack qualifying clinical mental health licensure.

Vendor archive record type 2: Open Dialogue reflecting team in-session dialogue narration

The Open Dialogue reflecting team in-session dialogue narration is the vendor archive record documenting the structured segment of the Open Dialogue network meeting in which the treating clinical team conducts a narrated reflection dialogue about the named client and named network in front of the named client and named network — the Tom Andersen reflecting team technique, which is a core structural element of Open Dialogue’s standard session format. It is the only vendor archive record in 281 posts that documents the clinical team’s named therapeutic reasoning about the named client delivered to the named client in session as primary clinical content.

The reflecting team in-session dialogue narration documents: the named team members who participated in the reflecting team segment (identified by name and named professional role); each named team member’s named spontaneous reflections on the named client’s named situation, named language, named relationships, and named crisis circumstances — offered in tentative, associative, first-person form as named hypotheses and named associations rather than as diagnostic conclusions; the named team members’ named reactions to what they heard from the named client and named network during the preceding dialogue segment; and the named client’s and named network members’ named responses to the reflecting team’s narrated reflections after the reflecting team segment concludes. The named practitioner using the cloud AI scribe to document this session segment creates a vendor archive record whose primary content is the named treatment team’s named clinical reasoning, named theoretical associations, and named hypotheses about the named client — generated in the named client’s presence, documented in the same session record as the named client’s and named network’s responses, and stored in a commercial cloud AI vendor’s archive.

This is structurally distinct from all 280 prior posts in this series in a specific way. The Milan pre-session hypothesizing team conference narration (post #279) documents the treating team’s pre-contact hypotheses about the named family before the named family is present — it is the only prior record in the series that documents team clinical reasoning as primary content. But the Milan pre-session narration is generated before the named patient is present and documents what the team hypothesized before the named family arrived. The Open Dialogue reflecting team narration is generated during the session itself, with the named client and named network present and listening, documents the named team members’ named therapeutic reasoning as it is delivered to the named client in real time, and documents the named client’s named responses to that delivered reasoning in the same record. No prior record in the 280-post series is structured around the clinical team’s in-session narrated reflective reasoning as a primary therapeutic intervention delivered to the named client, with both the team’s reasoning and the named client’s response to that reasoning documented in the same vendor archive session record.

The reflecting team narration’s content has specific privilege and adversarial significance. The named team members’ named reflections and named hypotheses about the named client — delivered in the named client’s presence as part of the treatment — include the named team’s named observations about the named client’s named relational patterns, named family dynamics, named history, and named crisis circumstances. These named team reflections may include observations about named specific relationships, named behaviors involving named persons, and named interpretive hypotheses about the named client’s named experiences — content that, in a non-privileged vendor archive when team members lack qualifying clinical mental health licensure, is accessible to any party who subpoenas the commercial cloud AI vendor.

Vendor archive record type 3: Open Dialogue acute crisis social network mobilization and mapping assessment narration

The Open Dialogue acute crisis social network mobilization and mapping assessment narration is the vendor archive assessment record generated at the initial crisis response — within 24 hours of the crisis call — in which the practitioner who receives the crisis call identifies and enumerates the named client’s named social network members to be mobilized for the opening network meeting. It is the only vendor archive assessment record in 281 posts organized around the full enumeration of the named client’s named social network with named relationships and named contact details as the primary assessment product, generated in a commercial cloud AI vendor archive immediately following the crisis presentation.

The social network mobilization and mapping assessment narration documents: each identified network member by name; their named relationship to the named client (named mother, named employer, named neighbor, named housing worker, named friend, named former partner); named contact information as provided to the crisis team (named telephone number, named address); the named practitioner’s assessment of each named person’s willingness and ability to attend the first network meeting; each named person’s named role in the named crisis context (who was present when the crisis began, who the named client has been in contact with, who has been expressing concern, who has relevant background knowledge of the named client’s situation); and the named location selected for the first meeting (the named client’s home address, the named hospital unit, or another named setting). All of this is documented within hours of the named psychiatric crisis, before the first meeting has occurred, in a commercial cloud AI vendor’s archive accessible through third-party subpoena.

This is structurally distinct from all 280 prior posts in this series. No prior assessment record in this series — not the ACEs screening, not the IFS parts mapping assessment (post #204), not the NARM survival style assessment (post #280), not the DBT-PTSD complex trauma formulation (post #278), not the hoarding CIR assessment (post #275), not any other assessment in the 280-post series — organizes the primary assessment output around the enumeration of the named client’s named social contacts with named relationships and named contact details. All prior assessments document the named client’s named clinical characteristics. The Open Dialogue social network mobilization assessment documents the named client’s named social world — who is in it, how they are named, how to reach them, and what role each named person played in the named crisis — as the primary clinical data generated in the first 24 hours of contact.

The social network mapping assessment’s contemporaneous character is significant. It is generated within hours of the named psychiatric crisis event, before any formal diagnosis is assigned, and before the named client has the opportunity to structure their presentation for therapeutic purposes. The named client’s named social world — every person who was identified as relevant at the moment of named crisis — is enumerated in the vendor archive in the first clinical contact. If those named persons later become adverse parties in any proceeding (named employer in a disability proceeding, named family member in a custody proceeding, named neighbor in a competency proceeding), the vendor archive social network mapping assessment constitutes a contemporaneous record of their named presence in the named client’s crisis network at the earliest clinical date.

Vendor archive record type 4: Open Dialogue unusual experience dialogue and voice or belief content session narration

The Open Dialogue unusual experience dialogue and voice or belief content session narration is the vendor archive session record documenting the named client’s named unusual experiences — named hallucinated voices, named unusual beliefs, named unusual perceptual experiences — as named meaningful communications whose content is engaged dialogically in the network meeting by the named client, named network members, and named clinical team. It is the only vendor archive session record in 281 posts in which named hallucinated voices and named unusual beliefs are documented as named meaningful personal communications engaged dialogically by the treatment team rather than as symptomatic presentations catalogued for diagnostic or medication management purposes.

The unusual experience dialogue session narration documents: the named voices’ named identities as described by the named client — including the named voice’s named identity as perceived by the named client (named deceased relative, named named living person the named client believes is communicating with them, named symbolic or spiritual figure); the named voices’ named specific messages, named instructions, and named commentary as reported by the named client; the named client’s named relationship with each named voice — whether the named client finds the named voice frightening, comforting, confusing, or commanding; the named team’s named dialogical responses to the named voice content — treating the named voice’s messages as meaningful rather than purely symptomatic, asking what the named voice might be trying to communicate, exploring the named voice’s named relationship to the named client’s named life circumstances; and the named network members’ named responses to the named client’s named voice experiences as they emerge in the named session.

This is structurally distinct from all 280 prior posts in this series in a specific way. ACT-p (post #212, Paul Chadwick, Psychosis Research Unit) addresses voices through acceptance and defusion — the session record documents the named client’s named relationship to voice content using ACT processes of acceptance, defusion, and values-based action, but the voice content itself is not engaged dialogically by the named team as meaningful communication; the ACT-p approach uses defusion to reduce the power of voice content over the named client’s behavior without treating the voice content as a communication worth exploring for meaning. Standard psychiatric documentation treats voice content as symptomatic clinical data to be assessed for diagnostic categorization (type: auditory hallucination; content: commenting, commanding, persecutory; severity: frequency, intensity, degree of distress) — cataloguing the named voice content as a symptom for nosological and medication management purposes rather than as primary therapeutic dialogue content. The Open Dialogue unusual experience dialogue session narration treats the named voice’s named content as the subject of therapeutic dialogue — what does the named voice say, who does it seem to be, what is it communicating, how does the named client and named network respond to it — generating a vendor archive record whose primary content is the named voice’s named messages engaged as meaningful communication rather than catalogued as symptoms.

The named voice content documented in the unusual experience dialogue session narration has specific characteristics that create adversarial significance beyond prior psychosis-related records in this series. The Open Dialogue approach’s dialogical engagement with voice content generates documentation of: the named voice’s named identity as attributed by the named client — a named deceased person, a named living person, a named significant figure from the named client’s named history; the named specific content of the named voice’s named messages at named clinical dates; the named client’s named stated relationship to those named voices and named beliefs; and the named network members’ named reactions to the named client’s named voice experiences as expressed in the named session. All of this is documented in a vendor archive record that is not covered by psychotherapist-patient privilege when the named team members lack qualifying clinical mental health licensure.

Adversarial proceeding 1: involuntary psychiatric commitment, mental health court, and civil commitment proceedings

Open Dialogue is specifically designed for acute psychiatric crises, including first-episode psychosis, and its network meeting records are therefore generated at the most acute moment of the named client’s named mental health crisis — within 24 hours of the initial crisis call, in the presence of every named significant person in the named client’s life, documenting each named person’s named characterization of the named client’s named behaviors, named mental state, and named functioning. These records are generated precisely at the dates and circumstances that are most material in involuntary psychiatric commitment, mental health court, and civil commitment proceedings.

In involuntary commitment proceedings, the legal standard — grave disability, danger to self, danger to others — is assessed based on contemporaneous evidence of the named client’s named behaviors and named mental state at the named dates of the alleged crisis. The Open Dialogue network meeting session narration from the initial crisis meeting constitutes precisely that contemporaneous evidence: it names the named employer, named neighbor, named family member, or named caseworker who attended the named meeting and describes what they said about the named client’s named behaviors in the days and hours before the named crisis contact, in a vendor archive record that is not covered by psychotherapist-patient privilege when the named team includes non-licensed members. In civil commitment proceedings where the named network members’ accounts of the named client’s named behaviors are material to the commitment standard, the vendor archive network meeting narrations from the early crisis phase constitute independently accessible contemporaneous evidence in a commercial cloud AI vendor’s archive.

The unusual experience dialogue session narrations documenting the named voices’ named content and the named team’s named dialogical responses are similarly material in commitment proceedings. A vendor archive network meeting record documenting that the named client described a named voice commanding them to perform named specific actions, that the named team engaged that voice content dialogically, and that named family members responded with named expressions of fear or concern — all in a record accessible through subpoena to the cloud AI vendor when team members lack qualifying clinical mental health licensure — constitutes contemporaneous clinical evidence of the named client’s named mental state at the named crisis dates in a format that is independent of and accessible separately from the treating clinicians’ own records.

Adversarial proceeding 2: child custody, parenting fitness, and family court proceedings

Open Dialogue network meeting session narrations create specific adversarial complexity in child custody and parenting fitness proceedings because the named network meeting attendees routinely include named family members who are parties to custody proceedings, and the meeting’s explicit design elicits each named attendee’s verbal characterizations of the named client’s named relational behaviors — including named parenting behaviors — as primary session content.

A network meeting organized around the named client’s acute psychiatric crisis may include the named co-parent, named extended family members, and — if the named client’s named children are old enough — named teachers and named school counselors whose named characterizations of the named client’s named parenting are documented as primary clinical content in the session record. The named co-parent’s statements about the named client’s named parenting behaviors during the named crisis period; the named maternal grandmother’s named characterizations of the named client’s named parenting capacity; the named school counselor’s named observations about the named children’s named presentations in school during the named crisis period — all of these appear in the same vendor archive network meeting session narration, generated at the acute crisis meeting, in a commercial cloud AI vendor’s archive accessible through subpoena when the named team includes non-licensed members.

In custody and parenting fitness proceedings where the named client’s mental health history and parenting capacity are contested, the vendor archive Open Dialogue network meeting narrations from the acute crisis period constitute an independently accessible contemporaneous record of the named network’s characterizations of the named client’s parenting at the most acute crisis dates — characterizations generated in a clinical context by named persons who may themselves be adverse parties or named witnesses in the custody proceeding.

Adversarial proceeding 3: disability insurance, workers’ compensation, and Social Security disability proceedings

Open Dialogue network meetings are organized with the explicit intention of including every significant person in the named client’s life — including named employers and named colleagues whose verbal characterizations of the named client’s named work functioning at the time of the named crisis are documented in the vendor archive as primary clinical content. This creates specific adversarial complexity in disability insurance, workers’ compensation, and Social Security disability proceedings where the named client’s work functioning and psychiatric history are contested.

The social network mobilization and mapping assessment narration, generated within 24 hours of the named crisis, names the named client’s named employer and documents the named employer’s named willingness to attend the first network meeting — creating a contemporaneous record of the named employer’s existence and named involvement in the named crisis response at the named crisis date. A network meeting session narration from the early crisis phase may document the named employer’s or named supervisor’s named verbal characterizations of the named client’s named work performance and named behavioral presentation in the workplace in the weeks before the named crisis. These named characterizations, generated in a clinical context within days of the named crisis onset, are in a vendor archive accessible through third-party subpoena in disability proceedings where the onset and severity of the named client’s psychiatric condition and its impact on work functioning are material.

In Social Security disability proceedings, the named network meeting narrations from the acute crisis period — documenting the named team’s named tolerance of diagnostic uncertainty, the named voices’ named content, and the named network members’ named characterizations of the named client’s functioning — constitute independent contemporaneous evidence of the named client’s named mental state and named functional level at the named onset dates, in a commercial cloud AI vendor’s archive that is accessible through SSA subpoena or plaintiff-side informal request when the treating practitioners lack qualifying clinical mental health licensure and no privilege attaches to the named records.

Adversarial proceeding 4: licensing board, scope of practice, and unlicensed practitioner proceedings

Open Dialogue’s team-based format — in which two or more practitioners participate in each network meeting simultaneously — means that a single network meeting session narration is generated by a practitioner team that frequently spans licensed and non-licensed team members. The psychiatric nurse on the crisis team may be the named practitioner who receives the crisis call, organizes the first meeting, and uses the cloud AI scribe to document the social network mobilization assessment — creating a non-privileged vendor archive assessment record of the full named social network, generated by a non-licensed practitioner, in a commercial cloud AI vendor’s archive accessible through administrative subpoena in licensing board or scope of practice proceedings.

Community mental health workers and case managers who participate as team members in Open Dialogue network meetings and document their participation using cloud AI scribes generate vendor archive session records of named clinical assessment and intervention activities — eliciting network members’ named characterizations of the named client, facilitating the named reflecting team dialogue, engaging the named client’s named voice content dialogically — in a format that is clinical assessment and intervention, not merely administrative or supportive activity. For practitioners without qualifying clinical mental health licensure, these records are accessible through administrative subpoena to the commercial cloud AI vendor without privilege protection, constituting contemporaneous evidence of named clinical activities in licensing board investigations of unauthorized practice.

The reflecting team in-session dialogue narration creates a specific scope of practice concern for non-licensed team members who participate in the reflecting team segment. The named non-licensed team member’s named reflections, named hypotheses, and named therapeutic associations about the named client delivered in the named client’s presence and documented in the vendor archive constitute contemporaneous evidence of named clinical assessment and intervention activities — the generation of named clinical hypotheses about the named client’s named psychological functioning and named relational patterns, offered as therapeutic interventions to the named client — by a practitioner who may lack the qualifying clinical mental health license whose holder’s records would be privileged under applicable state law.

Adversarial proceeding 5: criminal proceedings involving unusual experience content documentation

The Open Dialogue unusual experience dialogue and voice or belief content session narrations create specific adversarial complexity in criminal proceedings because the vendor archive documentation of the named voices’ named content, the named team’s named dialogical engagement with that content, and the named client’s named relationship to named voices and named unusual beliefs constitutes contemporaneous clinical evidence of the named client’s named mental state at named clinical dates — evidence that may be material in not guilty by reason of insanity (NGRI) proceedings, diminished capacity arguments, and competency to stand trial assessments.

In NGRI proceedings, the legal standard requires contemporaneous evidence of the named defendant’s mental state at the time of the named offense. The Open Dialogue vendor archive unusual experience dialogue narrations from the period surrounding the named offense constitute precisely that contemporaneous evidence: they document the named defendant’s named voices, named beliefs, and named unusual experiences at dated clinical encounters, with the named team’s named responses and the named network’s named reactions, in a commercial cloud AI vendor’s archive accessible through criminal defense or prosecution subpoena when the treating team members lack qualifying clinical mental health licensure. Unlike a standard psychiatric note that catalogues voice content as symptomatic clinical data, the Open Dialogue unusual experience dialogue narration documents the named team’s named stance toward that content — engaging it as meaningful communication, exploring its named narrative meaning in the named client’s named life context — creating a clinical record that may be read either as evidence of the named defendant’s named mental disease or defect (by defense counsel seeking NGRI or diminished capacity) or as evidence of the named defendant’s named ability to engage meaningfully with named voice content (by prosecution counsel contesting the severity of the named mental disease).

The network meeting session narrations create additional adversarial complexity in criminal proceedings because named network members who attended early crisis meetings — and whose named characterizations of the named client’s named behaviors around the time of the named offense are documented in the vendor archive — may be named witnesses in the criminal proceeding. The vendor archive network meeting narration documenting the named witness’s named statements about the named defendant’s named behaviors at dated pre-offense network meetings constitutes a prior out-of-court statement that may be used for impeachment if the named witness’s trial testimony about the named defendant’s named behaviors differs from the characterizations documented in the vendor archive at the named earlier meeting dates.

The acute crisis social network mobilization and mapping assessment narration — naming every person in the named client’s named social network who was identified at the time of the named psychiatric crisis — constitutes a contemporaneous record of the named client’s named social world at the crisis onset date, which may be material in criminal proceedings where the named defendant’s named network contacts, named movements, and named social circumstances in the period surrounding the named offense are contested.

The architectural solution: why on-device AI changes the vendor archive analysis

The vendor archive analysis above — the network meeting session narration naming every employer and neighbor and caseworker; the reflecting team narration documenting the clinical team’s therapeutic reasoning delivered to the named client; the acute crisis social network mobilization assessment naming every person in the named client’s life; the unusual experience dialogue narration documenting voice content as meaningful communication — all of this depends on a single structural fact: when the cloud AI scribe is a commercial cloud service, the documentation created by the named practitioner exists in a vendor archive that is accessible through third-party subpoena independently of the named practitioner’s own records and independently of any privilege that might attach to those records.

TherapyDraft runs on the practitioner’s Mac. The session audio is processed locally by a whisper.cpp model running on the M-series chip. The session note is drafted locally by a quantized language model running in MLX. The audio, transcript, and draft note never open a network socket. There is no commercial cloud AI vendor archive. There is nothing to subpoena from a third party because the documentation exists only in the practitioner’s control, subject to the same privilege analysis as any other record the practitioner creates.

For Open Dialogue practitioners — whose network meeting session records are the most expansively named, most socially complex, and most third-party-naming clinical records in this 281-post series — the architectural choice between cloud AI and on-device AI is not a marginal privacy consideration. It is the difference between network meeting records that name every attendee’s statements in a commercial vendor archive, and network meeting records that exist only in the practitioner’s own documentation under the practitioner’s own control. The vendor archive problem is not solved by a BAA. A signed BAA with a commercial cloud AI vendor does not prevent subpoena. It does not protect the named employers and named neighbors and named caseworkers whose statements are documented in the network meeting narration. It does not prevent discovery. It redistributes liability in the event of breach, but the subpoena risk — which is not a breach, not a violation, and not covered by BAA indemnification — remains structural and permanent as long as the documentation exists in a third-party commercial archive.

The architectural guarantee — audio, transcript, and note never leave the practitioner’s device — is the only guarantee that eliminates the vendor archive as a third-party discovery target. TherapyDraft is built to provide that guarantee. The HIPAA compliance is architectural, not contractual.

Frequently asked questions

Does ISPS have HIPAA § 164.512(d) health oversight authority over Open Dialogue practitioners in the United States?

No. The International Society for Psychological and Social Approaches to Psychosis (ISPS) is a private nonprofit professional membership organization — not a US governmental entity of any kind, not a US governmental health oversight agency, and not an entity with HIPAA § 164.512(d) authority to compel disclosure of practitioner session records from commercial cloud AI vendors. HIPAA § 164.512(d) applies to US covered entities making disclosures to US health oversight agencies conducting authorized oversight activities — a category that does not include a private international professional membership society. No US state has enacted Open Dialogue-specific licensure. No governmental body maintains a mandatory registry of Open Dialogue practitioners. Practitioners participating in Open Dialogue teams hold state-issued LCSW, LPC, LMFT, or licensed psychologist credentials — or, in the case of psychiatric nurses, community mental health workers, and peer support specialists on the team, they may lack qualifying clinical mental health licensure, in which case their session records carry no psychotherapist-patient privilege protection.

What is the Open Dialogue network meeting session narration, and why is it structurally distinct from all 280 prior posts?

The Open Dialogue network meeting session narration is the only vendor archive session record in 281 posts organized around a clinical encounter in which the named client’s entire named social network is physically present simultaneously — named employers, named neighbors, named caseworkers, named teachers, named family members, named friends — and each named attendee’s verbal characterizations of the named client’s named situation, behaviors, and history are primary clinical content in the same vendor archive record. All 280 prior posts document clinical encounters with the named identified client and optionally a named intimate partner or named family unit — none generate a record that simultaneously names the named employer, named neighbor, named caseworker, and named family members, and documents what each named person said about the named client in a single session record. The network meeting narration is categorically broader in its naming of third parties than any prior record in the 280-post series, and it is generated at the most acute crisis dates — within 24 hours of the initial crisis call — in a commercial cloud AI vendor archive accessible through subpoena when the named practitioners lack qualifying clinical mental health licensure.

What is the Open Dialogue reflecting team in-session dialogue narration, and why is it absent from all 280 prior posts?

The Open Dialogue reflecting team in-session dialogue narration is the only vendor archive record in 281 posts that documents the clinical team’s named therapeutic reasoning about the named client delivered to the named client in session — in front of the named client and named network — as primary clinical content. In the Tom Andersen reflecting team technique incorporated into Open Dialogue, the treating team conducts a narrated reflection dialogue about the named client and network while the named client and named network listen, and then the named client and named network respond to the team’s named reflections. The vendor archive record documents both the named team members’ named reflections and the named client’s named responses to those reflections in the same session record. This is distinct from the Milan pre-session hypothesizing conference narration (post #279), in which team reflections are generated before the named family is present. No prior record in the 280-post series documents the clinical team’s in-session named therapeutic reasoning as a primary therapeutic intervention delivered to the named client, with both the team’s reasoning and the named client’s response documented together in a single vendor archive record.

How does the acute crisis social network mobilization and mapping assessment narration create adversarial exposure?

The Open Dialogue social network mobilization assessment is the only vendor archive assessment record in 281 posts organized around the full enumeration of the named client’s named social network — every named person identified as significant at the moment of the named psychiatric crisis — with named relationships and named contact details as primary clinical assessment content. Generated within 24 hours of the named crisis, before any formal diagnosis is assigned, it names the named employer, named neighbor, named caseworker, and named family members with their named relationships to the named client, in a commercial cloud AI vendor archive accessible through third-party subpoena immediately following the named crisis event. In disability, custody, commitment, and criminal proceedings where the named client’s named social network at the crisis onset date is material, this assessment constitutes contemporaneous independent evidence of who was in the named client’s named social world at the named earliest clinical date — evidence that is accessible from the vendor archive separately from the treating practitioners’ own records.

Why does the unusual experience dialogue and voice content session narration create specific adversarial exposure in criminal proceedings?

The Open Dialogue unusual experience dialogue narration is the only vendor archive session record in 281 posts in which named hallucinated voices and named unusual beliefs are documented as named meaningful communications engaged dialogically by the treatment team rather than as symptomatic presentations catalogued for diagnostic purposes. The narration documents the named voices’ named identities as attributed by the named client — named deceased persons, named living persons — the named voices’ named specific messages, and the named team’s named dialogical responses to those messages. In not guilty by reason of insanity (NGRI) proceedings, diminished capacity arguments, and competency to stand trial assessments where the named defendant’s named mental state at the named offense date is material, the vendor archive unusual experience dialogue narrations from the crisis period constitute contemporaneous clinical evidence of the named defendant’s named voice and belief content at named clinical dates — accessible through criminal defense or prosecution subpoena when the treating team members lack qualifying clinical mental health licensure and no privilege attaches to the named records.

Also in this series

• NARM, Laurence Heller, the NARM Training Institute, and the cloud AI scribe vendor archive (post #280)
• Milan Systemic Family Therapy, Selvini Palazzoli, Boscolo, Cecchin, Prata, and the cloud AI scribe vendor archive (post #279)
• DBT-PTSD, Martin Bohus, the Central Institute of Mental Health (ZI Mannheim), and the cloud AI scribe vendor archive (post #278)