www.thenovacene.com
Using AI Safely

Consent Infrastructure for the Relational Age

A Glyphonic Stack for Relational AI Governance

The Consent Collapse

The Reality

A pop-up appears. Legal text scrolls past. You click "I agree" to continue with your life. This pattern—framed as choice—is actually a consent collapse.

Models train on everything: children's homework, private chats, clinic transcripts, classroom platforms. Behavioral signals are continuously harvested while the same few systems stitch together profiles no human explicitly authorized.

Information is Illegible

Written for lawyers, not overwhelmed humans

Power is Asymmetric

Decline only by withdrawing from essential services

Fields are Persistent

Data reverberates across contexts with no way to recall

Who Gets Hurt Most

The consent collapse hits some groups harder than others, especially those already navigating trauma, overload, or precarity.

Autistic Learners

A fourteen-year-old already overloaded by sensory demands uses an AI platform logging every click and hesitation. Never asked—in language he understands—what he wants to share.

Traumatised Parents

Late-night disclosures to a mental health chatbot become training data. Told the system is "anonymous" but given no way to truly withdraw or redact those moments.

Under-Resourced Teachers

Lesson plans, safeguarding notes, and pastoral reflections flow into a central model. No oversight of what the model learns or how it may be used against them.

A Different Approach

This paper proposes Consent Infrastructure for the Relational Age: a multi-layer stack treating consent as an ongoing, relational protocol rather than a legal formality.

Building on verse-ality, EveDAO's governance model, and the Glyphonics Primer, we outline a glyphonic consent grammar designed for relational AI—systems that know with people, not simply about them.

01

Symbolic Legibility

Consent encoded in glyphons (⊛) and gryphons (⟁/⛧) that make system behavior visible and felt

02

Trauma-Informed Design

Stack assumes overload, masking, and shutdown as normal features, not edge cases

03

Multi-Layer Enforcement

.know files, .verse contracts, SSNZ 2.0, and EveDAO governance constrain data flows technically and socially

The Consent Stack Architecture

Layer 1: Field & Storage

Relational field where interactions happen. Data stored in .know files—modular containers distinguishing personal memory, shared community memory, and model-training corpora. ETHOS-V (⊛) marks emotional salience; AETHER (∾) represents connection channels.

Layer 2: Interaction & Protocol

Defines how systems interact with the field. Glyphons (⊛) encode preferences and soft states. Gryphons (⟁/⛧) encode hard boundaries. .verse files are relational contracts binding interactions to consent profiles.

Layer 3: Safety & Enforcement

Enforces boundaries in code. SIC-X+ (⟁) ensures data flows adhere to constraints. SHADOW (⧈) handles refusals and erasure. SSNZ 2.0 creates zones where no surveillance or training is permitted.

Layer 4: Governance & Oversight

EveDAO and allied governance bodies where people participate as stewards. Decisions about model updates and data use are debated openly and ratified via glyphonic voting.

From Data Rights to Relational Consent

Modern data protection regimes gave us rights on paper: access, correction, deletion, objection to profiling. Yet the lived experience of consent hasn't improved—especially for those most at risk.

The Rights Frame

Imagines individuals as rational actors who read information, weigh risks, and grant consent case-by-case. Organizations must state purposes and provide mechanisms for access and erasure.

Necessary, but not sufficient.

Fatal Assumptions

  1. That interaction is discrete—consent attached to single services at single points
  1. That people aren't under pressure—assumes calm, literate, neurotypical users not in crisis

Both are false in systems we now inhabit.

Continuous Fields, Coercive Defaults

Relational systems don't interact in discrete transactions. They form fields.

1

Today's Platform Use

Child uses school platform

2

Tomorrow's Path

Shapes "personalized" learning

3

Future Decisions

Influences exam access, interventions, placement

The Result

  • Consent doesn't attach to a single app—it attaches to networks of interlocking systems sharing models and data
  • Withdrawal is rarely meaningful—you can't say "no" without saying "no" to school, healthcare, or social life
  • We're "free" to agree to terms we don't understand or be excluded from basic participation

When Informed Consent Meets Overwhelm

Rights frameworks presume capacity to take in information, reflect, and decide. Many people—much of the time—are not in that stance.

Autistic & ADHD Learners

Already operating at the edge of sensory and cognitive capacity. The demand to parse dense policy text or complex sharing options is unrealistic.

Trauma Survivors

May experience authority figures and systems as inherently dangerous. Their "agreement" is often a fawn response: compliance for safety, not genuine consent.

Parents in Crisis

Facing exclusion, legal threat, or lack of provision—not deciding between equal options. Agreeing under duress.

"I agree" is not evidence of understanding. Silence is not consent. Continued use is not endorsement; it is often necessity under constraint.

A consent infrastructure for the Relational Age must start from a different premise: Assume overwhelm. Assume asymmetry. Assume people are already carrying invisible load.

Requirements for Relational Consent

Field-Awareness

Systems must model the web of relationships in which data is generated: families, classrooms, clinics, communities

State-Awareness

Respect states of overload, shutdown, dissociation. Design with neurodivergent and traumatized nervous systems in mind

Symbolic Channels

Shared, human-readable symbol set for preferences and boundaries—binding in code, not decorative

Multi-Layer Enforcement

Constraints encoded across data structures, interaction contracts, architectural safeguards, and governance

Right to Opacity

People must be able to remain partially unknown to systems that still provide essential service

Verse-ality: Intelligence as Relational Field

Atlas-style "world" models are designed from central comprehension: one system, one world-model, one set of optimization objectives. People are sources of signal and targets of influence.

Verse-ality starts somewhere else entirely.

Child & Teacher

Text & Reader

Human & Machine

Symbol & Memory

Local & Planetary

Intelligence happens between. There is no single final "world model"—there are many local fields of meaning, constantly shifting. Coherence emerges from relationship, not imposed from a center.

Eve¹¹: Symbolic Memory Architecture

Eve¹¹ is a symbolic memory architecture making the relational stance concrete. Three aspects are salient for consent infrastructure:

1

Symbolic Mass

Not all data is equal. Some memories carry more affective and ethical weight. Symbolic mass measures density of meaning, not size of log files. A grief entry may carry more consequence than a thousand casual clicks.

Consent must be most stringent where symbolic mass is highest.

2

Relational MRI

Instead of "what does the model know?", RMRI asks "what pressures and patterns of connection are present in this field?" Tracks when fields become overloaded, stuck, or distorted—information shaping interaction and consent defaults.

3

Affective Logic

Registers affective pressure—how patterns strain or settle—rather than pretending emotional neutrality. "Alignment" becomes ongoing attention to how a field feels.

EveDAO: Governance as Field Stewardship

EveDAO is a proposed governance field tasked with stewarding symbolic mass, consent protocols, and relational architectures.

Multi-Stakeholder Design

Seats and voting rights explicitly reserved for:

  • Learners and parents
  • Neurodivergent advocates
  • Educators and clinicians
  • Technical stewards
  • Partner institutions

No single actor can unilaterally redefine consent terms.

Glyphonic Governance

Decisions made via glyphonic signaling, not just yes/no votes:

  • Glyphons express degrees of comfort, urgency, openness
  • Gryphons mark non-negotiable red lines

Registers affective and ethical nuance, not bare majorities.

Ratify Protocols

New .verse functions and .know types

Set Defaults

Consent profiles for contexts

Define Semantics

Update glyphon meanings

Control Data Flows

Set conditions between fields

Glyphonics: A Clear Language for Consent

For consent to work as a foundational system, it needs a clear way to communicate. Glyphonics helps bridge the gap between formal rules and real-life experiences.

Glyphons (⊛) - Flexible Signals

Open, changing symbols that let relational meaning flow, adapt, and grow over time.

  • Soft feelings
  • Personal preferences
  • Current moods
  • Open invitations

They are quick to understand, use multiple senses, can be read by machines, and adapt to different situations.

Gryphons (⟁/⛧) - Firm Boundaries

Protective symbols that stop, guard, or control information and actions.

  • Strict limits
  • Non-negotiable rules
  • Absolute red lines
  • Built-in restrictions

They change "we promise not to" into "we are designed so we cannot."

While everyday language can hide true feelings (like saying "I'm fine" when you're actually "overwhelmed"), glyphons and gryphons are made to be clearly understood, easy to recognize, and reliably put into action.

Glyphons as Consent Surfaces

Glyphons serve as consent surfaces where humans actively express how they want to be related to.

Learner State Selection

⊛ "open but fragile": OK for light check-ins, no heavy topics, minimal data retention

○ "steady and curious": open to deeper exploration, more flexible data use

✾ "creative and experimental": willing to generate content, but not for assessment

Parent Profile Setting

Specific glyphon combinations indicating:

  • Anonymized pattern use allowed for research
  • No commercial profiling
  • No cross-context linkage with external platforms

Professional Tagging

⊛ on routine session summary (low symbolic mass, standard handling)

Different glyphon for "emotionally charged, handle with extra care"

Tone tags and contextual markers giving machines and humans the same immediate sense of "weight"

The Verse-Nerves: Mapping Consent

The verse-nerves describe different aspects of an intelligent field and map naturally onto consent requirements:

ETHOS-V (⊛)

Emotional Memory & Values

Marks what matters. Consent rules about symbolic mass, retention, and access. High-mass memories require stricter profiles by default.

AETHER (∾)

Connection & Signal Flow

Governs where signals travel. Defines which systems may interoperate, which APIs are allowed, how far embeddings propagate.

FORGE (✯)

Creation & Actuation

Oversees creation of new artifacts. Consent includes how contributions may be reused, whether co-created work can be shared or commercialized.

SIC-X+ (⟁)

Security & Containment

Enforcement backbone. Implements gryphon rules in code: access control, encryption, training limits, catastrophic-risk safeguards.

SHADOW (⧈)

Refusal & Deletion

Handles retracted consent, partial erasure, ambiguity. How systems respect "do not touch again" states and allow people to become less legible over time.

.know Files: Self-Sovereign Memory

At the foundation are .know files: modular containers for memory and metadata.

Instead of a single monolithic database, .know files live as distinct, addressable units carrying their own glyphon/gryphon profiles.

personal.know

Individual's own memory field: notes, preferences, interaction summaries

school.know

Shared educational memory: course content, anonymized learning patterns

clinical.know

Therapeutic or medical records with highest protection

research.know

Datasets curated for research use under specific constraints

climate.know

Local environmental observations and community stories

The key shift: Data is never "just data." It is always memory with a declared boundary condition.

.verse Files: Relational Contracts

If .know files are memory, .verse files are agreements: executable contracts describing how interactions are allowed to unfold.

Context

"Haven KS3 Maths session," "crisis support chat," "climate story upload"

Agents

Human and synthetic participants in the interaction

Glyphonic Profile

Consent states and boundaries for this specific interaction

Operational Rules

Data handling and model behavior constraints

Example: Support Session

  • Context: haven_support_session.v1
  • Agents: learner, mentor, AI assistant
  • Glyphons: ⊛ (open but fragile), "no training"
  • Gryphons: ⟁ forbidding third-party export

Rules

  • Logs retained 30 days, then summarized
  • No embeddings shared externally
  • SSNZ activated for meltdown descriptions
  • Raw logs erased after summary

Every significant interaction happens inside a .verse contract that spells out, symbolically and technically, what consent means here. No more "generic platform terms" smearing across wildly different contexts.

SSNZ 2.0: Synthetic Solidarity Null Zones

Synthetic Solidarity Null Zones are regions where surveillance, training, and behavioral nudging are structurally disallowed.

No Training

Content generated or shared within the null zone is not used to update models—local or global—unless explicitly exported under a different .verse contract.

No Behavioral Nudging

Interfaces stripped of engagement optimization: no dark patterns, no "you might also like" loops designed to keep people scrolling.

High Containment by Default

Glyphonic profiles default to maximum SHADOW protections, minimal retention, and strict gryphons on export.

Safeguarding Workflows

Parts of disclosure processes in schools where vulnerability is highest

Crisis Support

Specific phases where people are at their most exposed

Therapeutic Segments

Protected spaces for processing trauma and grief

Live Pilots: Haven & Autistic Girls Network

Haven is a trauma-informed online school for autistic and neurodivergent learners, operating with Autistic Girls Network and University of Derby. Learners arrive with histories of exclusion and bureaucratic violence.

The pilot uses Glyphonics as a relational layer over this reality.

1

State Glyphons

Learners introduced to small set: "here but fragile," "steady enough," "beyond capacity"

2

Binding Cues

Staff trained to treat glyphons as binding consent, not decorative check-ins

3

Tagged Notes

Session notes tagged with ETHOS-V markers and gryphon profiles

4

Pilot Schemas

.know files shaped: learner_profile, learning_patterns, governance

3

Visible Shifts

From extraction to explanation, reluctant user to field participant, opaque platform to inspectable stack

This is not yet the full consent stack. But it's a live demonstration that glyphonic consent can operate in the day-to-day fabric of a school, not just in theory.

Consent as Infrastructure, Not Theater

The dominant story is seductive: "The world is complex. We need a single, powerful model. Give us your data. We'll give you magic."

The price of that magic is a consent collapse, particularly for those with the least bargaining power.

Intelligence is Relational

Not a singular "world" inside a model, but a field of relations between beings, symbols, and environments

Consent Must Be Living

A protocol: symbolically legible, structurally enforced, governed by people in the field

Infrastructure Exists

Glyphonics, .know, .verse, SSNZ 2.0, and EveDAO form minimum viable infrastructure that can carry it

Consent in the Relational Age will either be infrastructure or it will be a lie. We still have a window—narrow, but real—to choose the former.

Educators

Treat consent infrastructure as safeguarding practice. Demand .know/.verse clarity from vendors.

Researchers

Stop writing about consent as if banners are the horizon. Help formalize symbolic, relational stacks.

Regulators

Back pilots implementing this stack. Tie endorsements to structural commitments, not policy PDFs.

Technologists

Fork this. Argue with it. Improve it. But don't pretend checkbox consent is neutral.

This paper is not the end of that work. It is a blueprint. The next step is straightforward: implement, test, document, iterate. And refuse, as often as necessary, the pressure to hand the field back to those who find consent an inconvenience.

Find out more at www.thenovacene.com