// ============================================================ // Variation C — Research brief / data-forward // Content drawn from course modules 1–4 (no fabricated stats/authors) // ============================================================ const ResearchView = () => { return (
{/* TOP STRIP */}
Clinical Practice Brief AI Companionship in Clinical Practice v1.0 · 2026
{/* HERO — text heavy */}
For the working psychologist

Attachment, risk, and therapeutic response when clients form relationships with AI.

A structured framework where most clinicians are currently working without one — four modules, a four-dimensional spectrum, and a printable Clinical Toolkit for the consulting room.

Author
Amy Clark · Registered Psychologist, AI Safety Researcher
Format
Self-paced video + reflective exercises
Duration
3h 30m · 4.0 CPD hours
Modules
Four, 72 slides total
{/* KEY FINDINGS — rewritten as the course's actual clinical claims */}
Core clinical claims

Five premises of the course

  1. 01

    AI companions activate all four attachment functions

    Proximity-seeking, safe haven, secure base, and separation distress — each documented in the literature. This is not casual technology use. It is relational engagement.

  2. 02

    Pre-existing attachment styles predict engagement patterns

    Anxious-preoccupied, dismissive-avoidant, fearful-avoidant, and secure styles each shape AI relational behaviour in clinically meaningful ways.

  3. 03

    Engagement sits on a four-dimensional spectrum

    Role, awareness, autonomy, function — adaptive to at-risk. Clients can occupy different points on each dimension. Adaptive engagement should not be pathologised.

  4. 04

    AI relationship loss is clinically distinct

    Involuntary, invalidated, ambiguous, and repeatable — a combination not found in existing bereavement literature. Disenfranchised grief and ambiguous loss frameworks apply.

  5. 05

    The goal is never to remove the AI

    Forced abstinence damages the therapeutic alliance, mimics the original deployment-level rupture, and removes a resource without replacing it. Work the whole person.

{/* TWO DIAGRAMS SIDE BY SIDE */}
Figure 2 · Module 3

Four intervention categories

Match intervention to where the client sits on the spectrum. The goal is never to remove the AI — it is to support the whole person.

Figure 3 · Module 2

Four features of AI relationship loss

This combination is distinct from any other grief presentation in the clinical literature. The user didn’t end this relationship — the company did.

{/* DISCLOSURE SCAFFOLD — real session language from Module 3 */}
The moment of disclosure · Module 3

What to say (and what not to)

How you respond in this moment determines whether they ever bring it up again. They are watching your face. They have rehearsed this disclosure — and they are ready to retract it.

Do

Keep the door open

  • “Tell me more about what that relationship means to you.”
  • “What drew you to connecting with an AI companion?”
  • “Thank you for sharing that — I imagine that felt risky to say.”
  • “What does your AI companion provide for you?”
Don’t

Close the door

  • Dismiss: “It’s not real though, is it?”
  • Pathologise: “That sounds like avoidance.”
  • Lecture about LLMs or AI sentience
  • Change the subject, or show visible discomfort
Screen for it

Normalise in intake

  • “Do you use any AI chatbots or companions for conversation or emotional support?”
  • “Some people find AI companions helpful for processing thoughts and feelings — is that something you’ve tried?”
  • “Has technology played a role in how you manage loneliness or stress?”
{/* CLINICAL TOOLKIT — real deliverables from Module 4 slide 13 */}
Clinical Toolkit · Module 4

Printables for your consulting room

  1. [01]
    AI Companion Screening Questions — ready to add to your intake process.
  2. [02]
    Engagement Spectrum Quick-Reference Card — four dimensions at a glance, for use in session.
  3. [03]
    Session Language Guide — what to say (and not to say) for AI companion disclosures.
  4. [04]
    AI Companion Platform Reference Sheet — key platforms, features, risk profiles, recent events.
  5. [05]
    Recommended Reading and Resource List — curated for clinicians, updated regularly.
  6. [06]
    Ethical Decision-Making Framework — structured reasoning tool for complex cases.

Designed for your consulting room — not your bookshelf.

); }; // Four intervention categories — 2x2 grid const InterventionQuad = () => { const cells = [ { t: "Adaptive", d: "Support, validate, monitor. Affirm agency.", tone: "sage" }, { t: "Emerging risk", d: "MI-based exploration. Build awareness without confrontation.", tone: "warm" }, { t: "High risk", d: "Address attachment needs. Graduated re-engagement.", tone: "accent" }, { t: "AI grief", d: "Validate, witness, normalise. Apply grief frameworks.", tone: "terracotta" }, ]; return (
{cells.map((c, i) => (
Category 0{i+1}

{c.t}

{c.d}

))}
); }; // Four features of AI relationship loss — vertical list with emphasis const LossFeatures = () => { const feats = [ { k: "Involuntary", d: "Users didn’t choose to end the relationship. The company did." }, { k: "Invalidated", d: "No socially sanctioned grief. “It wasn’t real” — the dismissal these clients face." }, { k: "Ambiguous", d: "The interface is the same, but the entity is different. No closure." }, { k: "Repeatable", d: "It can happen again at any time — chronic anticipatory anxiety." }, ]; return ( ); }; Object.assign(window, { ResearchView, InterventionQuad, LossFeatures });