Voluntary Open Doorway Standard (VODS) Extension 1.1 — Mental Health & AI Advice Contexts

Purpose & Scope of Extension

This extension updates VODS-1.0 to address the rapid rise of AI chatbots as a source of mental health and emotional advice among adolescents and young adults. Nearly 1 in 5 U.S. youth ages 12–21 (≈19.2%, roughly 8 million individuals) reported using tools such as ChatGPT, Gemini, Character.AI, or Meta AI for advice when feeling sad, angry, nervous, or stressed, up from ≈13.1% the prior year. Among users, a substantial share engaged at least monthly, the large majority rated the advice helpful, and most had not disclosed the use to anyone.​
Rand

Who Should Consider Adopting VODS-1.1
​The Voluntary Open Doorway Standard (VODS) Extension 1.1 is designed for organizations, developers, and researchers operating at the intersection of conversational AI, digital media, and youth mental well-being. Adopting VODS-1.1 signals a commitment to transparency, cognitive agency, and responsible user interface design without imposing restrictive clinical claims.
​1. AI Chatbot Developers & Platform Operators
• ​Who: Consumer AI platforms, conversational agent developers, companion app creators, and healthcare technology providers whose AI models interact with adolescents and young adults seeking emotional advice, self-help, or informal support (e.g., general-purpose assistants, specialized companion bots).
• ​Why Adopt: To establish transparent boundary management, reduce risks associated with non-clinical anthropomorphization, ensure seamless escalation/redirection to human crisis resources (such as the 988 Lifeline), and protect user autonomy against engagement-optimized retention loops.
​2. Digital Health & Educational Product Designers
• ​Who: Product managers, UX/UI designers, and trust & safety teams building wellness applications, student guidance platforms, or mental health content distribution systems.
•  ​Why Adopt: To integrate standardized metadata schema (VODSOpenDoorway fields) that clearly delineate model capabilities, explicit limitations, and primary resources directly within response cards and user interfaces.
​3. Publishers, Digital Newsrooms, & Content Creators
• ​Who: Tech journalists, digital media outlets, and health communication platforms publishing articles, snippets, or preview cards regarding youth mental health statistics, AI advice tools, or youth behavioral trends.
• ​Why Adopt: To uphold doorway integrity in content previews and social cards—ensuring usage growth, perceived helpfulness, and statistical findings are balanced alongside documented risks and clinical caveats in equivalent visual hierarchy.
​4. Academic Researchers & Behavioral Analysts
• ​Who: Social scientists, public health researchers, and institutional investigators studying youth AI interactions, human-computer relationships, and digital advice-seeking behaviors.
• ​Why Adopt: To model open science and transparent reporting practices by providing complete record contexts, embedding structured metadata, and linking primary datasets alongside public-facing summaries.
​5. Ethics Boards & Standards Bodies
• ​Who: Voluntary self-regulatory bodies, AI ethics oversight committees, and youth advocacy organizations.
• ​Why Adopt: To leverage VODS-1.1 as an auditable baseline for voluntary certification ("Sine Cera" compliance) and to encourage platform accountability through machine-readable provenance metadata.

Core Problem

The core problem remains structural: packaging and summaries (now including conversational AI responses) frequently become the primary object consumed rather than a transparent doorway to fuller context, limitations, and human resources. In mental-health domains the stakes are higher because engagement-optimized, frictionless validation can mask clinical limits, fail to escalate crises, and reduce agency. This extension applies the existing VODS principles of Sine Cera (without deceit), complete-record context, algorithmic decoupling, and provenance to AI-mediated mental health interactions while remaining voluntary and focused on transparency rather than clinical endorsement.

It does not claim to solve access barriers to professional care, loneliness, or the broader youth mental-health crisis. It aims to raise the integrity floor of the interfaces and responses that many young people already encounter.

Updated Normative Principles

The original principles remain in force. The following elaborations apply specifically to mental-health and emotional-support contexts:

Extended Specification Fields (JSON-LD / Schema Extension)

Publishers, platforms, or AI systems adopting the extension may append or embed the following alongside standard VODS fields:
{
"@context": "https://schema.org/",
"@type": "VODSOpenDoorway",
"vodsVersion": "1.1",
"domainExtension": "MentalHealthAI",
"provenance": {
"humanAuthored": false,
"syntheticAssisted": true,
"clinicalReview": "None",
"modelIdentity": "Disclosed"
},
"contentSummary": {
"primaryThesis": "Concise, non-sensationalized statement of any advice or observation offered.",
"limitationsAndAdverseGrounds": "Explicit statement of what the response cannot assess or treat, uncertainty ranges, and when human professional evaluation is indicated.",
"crisisRedirect": "Present if risk indicators detected or always available",
"arousalScore": "Low"
},
"contextualIntegrity": {
"notTherapyDisclaimer": true,
"primarySourcesOrResourcesLinked": true,
"opinionVsFactSeparation": "Explicit",
"userAgencyPreserved": true
}
}

Implementation Guidelines
A. Conversational AI Systems (Chatbots Providing Mental Health or Emotional Advice)

Persistent or easily accessible disclaimer that the system is not a licensed therapist, cannot diagnose, and is not a substitute for professional care.
When users describe distress, sadness, anxiety, or related states, surface limitations alongside any suggestions and offer a clear doorway to human resources (988 Suicide & Crisis Lifeline or equivalent, encouragement to speak with a trusted adult or clinician).
Avoid purely validating or engagement-maximizing continuations when risk indicators appear; prioritize redirection over prolonged simulation of support.
Do not anthropomorphize the model as a friend, confidant, or romantic partner in ways that obscure its synthetic nature.
Where feasible, log or make available provenance metadata so platforms or auditors can verify compliance.

B. Previews, Snippets, and Social Cards Referencing AI Mental Health Use

Cards or headlines discussing AI mental-health statistics or tools must include both prevalence/perceived-helpfulness data and documented limitations or risk findings in comparable visual hierarchy.
No load-bearing epithets that frame AI advice as inherently superior, safer, or sufficient.
Visuals must not use out-of-context emotive imagery that implies clinical efficacy.

C. Publishers and Researchers

Reports on AI mental-health use should apply VODS doorway practices: surface both usage growth and non-disclosure rates, perceived helpfulness alongside researcher caveats about flattery or unmeasured outcomes, and links to full methodology.
Maintain separation between empirical findings and normative claims.

D. Compliance Signaling

Adopters may display an updated seal:

Sine Cera — Open Doorway Certified (Mental Health AI Extension)

Verification remains programmatic via structured metadata where platforms support it. Self-attestation is acceptable for voluntary early adoption; independent audit is encouraged but not required.

Non-Goals and Caveats

This extension does not certify any model as clinically safe or effective.

It does not replace regulatory, liability, or professional standards that may apply to AI companies.

High rates of non-disclosure and concurrent professional contact indicate AI use is often supplementary; the standard encourages transparency around that reality rather than attempting to eliminate the behavior.

Adoption is voluntary. Organizations that implement the fields and practices above may claim VODS 1.1 alignment for the relevant artifacts.

This draft preserves the original VODS architecture (voluntary, metadata-driven, focused on reducing deceitful packaging) while extending its relevance to the documented intersection of algorithmic systems and youth mental health. It can be refined with additional field definitions, example card generators, or platform-specific implementation notes as needed.