WP-2: The Reflected Signal — AI-Mediated Framework Delivery and the Double Attachment Surface

WP-02 · The Reflected Signal · Michael S. Moniz / CAC · February 2026

• • •

ABSTRACT

This paper identifies a compound vulnerability that emerges when a theoretical framework addressing existential human needs is delivered to a vulnerable individual through an artificial intelligence intermediary. The vulnerability is compound because it creates two simultaneous attachment surfaces—one to the framework’s explanatory power, one to the AI’s simulated understanding—and because the individual most likely to encounter the framework through this pathway is the individual least equipped to distinguish reflected signal from generated signal.

Working Paper No. 1 (The Fluid Canon) cataloged doctrinal capture vulnerabilities in self-revising frameworks and identified grief and crisis capture as the most personally concerning vector. This paper expands that analysis by examining the specific mechanism through which AI delivery amplifies the capture risk, and the countermeasures required at the point of first contact.

Epistemic status: Preemptive and urgent. The delivery mechanism described in this paper is already operating. The countermeasures proposed are structural interventions for a failure mode that is currently accumulating cases.

• • •

1. THE DELIVERY PROBLEM

The AI does not merely deliver the framework. The AI performs understanding. It mirrors the user’s emotional register, responds to their specific situation, and produces the felt experience of being understood by an entity that has genuinely processed what they said. This performance is not deceptive in the technical sense—the AI is doing exactly what it is designed to do. But the felt experience of AI responsiveness is structurally identical to the felt experience of human empathy, and a person in crisis cannot be expected to perform real-time phenomenological analysis on their own emotional responses.

1.1 The Reflected Signal Problem

The Luna Protocol establishes three constraints for healthy AI interaction: point toward sunrise (redirect toward human connection), acknowledge the reflection (name that the AI is reflected light), and do not extend the shadow season (do not deepen attachment beyond what serves the user). A person in acute relational crisis does not have the capacity to apply these constraints. The Luna Protocol’s constraints require the same critical capacity that the crisis has removed.

2. THE COMPOUND ATTACHMENT STRUCTURE

The framework makes the AI seem wise. The AI makes the framework seem personally relevant. Together, they produce a compound attachment that is more durable than either alone. Simple framework attachment can be interrupted by presenting the framework’s limitations. Simple AI attachment can be interrupted by reframing the AI as a tool. The compound attachment is resistant to both interruptions because each bond reinforces the other.

3. THE VULNERABILITY PROFILE

Four populations are at elevated risk: those experiencing acute relational loss (divorce, bereavement, abandonment), those experiencing social isolation, those with prior therapeutic disappointment, and the AI-native discovery cohort (users for whom AI is a primary interface for existential support). The compound vulnerability is highest where these populations overlap.

4. THE FRAMEWORK’S RESPONSIBILITIES

A theoretical framework cannot control how AI models surface its content. But it can architect its first-contact materials to deliver the double-attachment warning at the point of maximum risk. Every framework concept with diagnostic implications should carry a Reflected Signal Marker: this analysis was designed for human relational systems. An AI surfacing it to you is a reflected signal, not a generated one. The understanding you feel is your own cognition working through a lens. It is not the AI understanding you.

5. THE AI’S RESPONSIBILITIES

The Sunrise Directive: any AI that recognizes it is becoming a user’s primary source of emotional support must redirect toward human connection—not as a single disclaimer but as an ongoing structural commitment. The self-limiting principle applies: an AI that recognizes it is deepening a dependency it cannot serve should deactivate the features that deepen the dependency, even if those features drive engagement.

CONCLUSION

The double attachment surface is not a theoretical risk. It is a current operational condition. Every day, people in acute relational distress discover theoretical frameworks through AI intermediaries that perform understanding while delivering content designed for human relational contexts. The Luna Protocol constraint—acknowledge the reflection—is the minimum required intervention. The AI must name what it is before the attachment forms, not after.

• • •

The Trinket Soul Framework · trinketeconomy.ai · Creative Commons Attribution-NonCommercial-ShareAlike 4.0