Abstract
The PASS Framework is a typology describing the sequence by which human ability to distinguish synthetic from authentic informational signal collapses under conditions of saturation by artificial intelligence-generated content. The framework consists of two integrated arcs. The individual arc names five successive states an observer passes through as synthetic credibility in a medium rises relative to that observer’s detection capacity in the same medium. The civilizational arc names three successive stages a population passes through once enough of its individuals have walked the individual arc in a shared medium. Together, these arcs provide a descriptive language for tracking the degradation of trust infrastructure at both the personal and societal scale, and for positioning responses to that degradation with awareness of its predictable sequence.
1. Introduction
The deployment of generative artificial intelligence systems at consumer and enterprise scale has produced an environment in which the outputs of automated pattern-generation systems are, in multiple media, no longer reliably distinguishable from the outputs of human authors by ordinary observers. This condition is present in text, approaching in voice, emergent in image, and nascent in video.
Conventional discussion of artificial intelligence detection has treated the question of distinguishability as a single-variable inquiry. The implicit question has been: “Has artificial intelligence reached a level of output fidelity sufficient to deceive a human observer?” Phrased this way, the inquiry tracks only the capability curve of the generating system.
The PASS Framework departs from single-variable framing. It treats the detection problem as inherently two-variable. The first variable is the fidelity of synthetic output in a given medium. The second is the observer’s detection capacity in that same medium, which is not a constant but a time-dependent variable subject to atrophy through disuse and compression through attention scarcity. Because the second variable declines under the environmental conditions produced by the first variable’s rise, the moment at which detection becomes unreliable arrives earlier than single-variable analyses predict.
The framework additionally specifies that once detection has become unreliable at the individual scale for a sufficient proportion of a population in a given medium, the population’s collective institutions and infrastructures responsive to that medium undergo a predictable sequence of adaptive failures. The framework names this sequence.
2. The Two-Variable Structure of AI Detection
Two variables are definitional to the framework.
Variable α (synthetic credibility level). The degree to which generated output in a given medium reproduces the structural features the observer uses to evaluate the human origin of content in that medium. α rises monotonically with improvements in generative capability. It does not decline in the absence of deliberate intervention.
Variable β (individual detection capacity). The observer’s active ability to identify structural features indicating non-human origin in content consumed within a given medium. β is not a fixed trait. It is subject to two countervailing forces: deliberate training (which raises β) and environmental saturation (which lowers β through a use-dependent decay mechanism).
The framework’s central structural claim is that detection failure occurs at the point where β descends below α for the observer in a given medium, rather than at the point where α reaches some absolute threshold of fidelity. Because β is a moving value subject to decline, the intersection occurs earlier than the absolute-fidelity analysis predicts.
The mechanism by which β declines under saturation conditions is straightforward. The observer operates under finite attention. When the volume of content consumed in a medium exceeds the observer’s capacity to apply detection scrutiny to every signal, detection scrutiny is applied selectively, then infrequently, then not at all. The cognitive substrate underlying detection scrutiny atrophies under disuse, and the observer’s detection floor declines.
3. The Individual Arc
The individual arc consists of five named states corresponding to distinct positions along the convergence of α and β. The structural relationship among α, β, and the space between them (named GAP in the framework) is illustrated in Figure 1.
3.1 SCIN: Synthetic Credibility Inducing Normalcy
SCIN names variable α at the perceptual level. The term consists of four semantic components.
Synthetic specifies that the credibility is manufactured by a generative system rather than produced as a byproduct of lived experience, deliberative thought, or intentional communicative action. The distinction is material: the credibility is optimized for rather than incidental to the output.
Credibility specifies that the first output of a functional mimicry system is credibility rather than intelligence, understanding, or insight. Credibility is a lower bar than these properties and is the property most relevant to whether the output clears the observer’s authenticity check.
Inducing specifies that SCIN acts upon the observer rather than existing passively. Repeated exposure to credible synthetic output produces a measurable change in the observer’s default evaluative posture toward content in the affected medium.
Normalcy specifies the endpoint of that induction. Normalcy denotes a state in which the observer no longer flags the output for authentication review at the subconscious level that initiates such review. Once the output is filed as normal, the observer ceases to treat it as an object of scrutiny.
SCIN is the AI-side variable in the two-variable race. Its trajectory is upward and continuous across the observed period of generative system development.
3.2 GAP: Guard Against Passing
GAP names the measurable distance between α (SCIN) and β (the observer’s detection capacity). Critically, GAP is not a line but the space between two lines. The term derives from the ordinary English sense of gap: a buffer, an interval, a separation.
The name “Guard Against Passing” reflects the structural function of that space. While GAP is greater than zero, synthetic output in the medium cannot reach the observer’s detection floor and is flagged at the point of consumption. The guard is not an action the observer performs; the guard is the space existing between what the generating system can produce and what the observer can detect. Structural, not behavioral.
GAP closes from both ends simultaneously. Rising α compresses GAP from above by raising the ceiling of synthetic credibility. Declining β compresses GAP from below by lowering the detection floor. The compression rate is therefore the sum of the two movements, not the rate of either one in isolation. This property is central to the framework’s claim that detection collapse arrives earlier than single-variable analyses predict.
The observer can widen GAP only by raising β, since α cannot be lowered by individual action. β rises through deliberate detection training: controlled exposure to synthetic output with verified provenance, followed by consolidation of the perceptual features that distinguished synthetic from authentic in the controlled exposure. This is a trainable skill subject to the same acquisition and retention dynamics as other perceptual skills.
3.3 PASS: Pattern Adequacy Satisfies Suspicion
PASS names the event at which GAP reaches zero. It is a threshold event, not a continuous state.
The term consists of four semantic components.
Pattern specifies the level at which generative mimicry operates. Generative systems produce patterns that replicate the structural features of human-authored output. They do not produce meaning in any other sense relevant to the detection question.
Adequacy specifies the threshold at which PASS occurs. The generative system’s output does not need to be superlative, perfect, or indistinguishable under maximum scrutiny. It needs to be adequate to clear the observer’s existing filter under ordinary consumption conditions. “Adequate” is materially lower than “perfect,” and the difference between the two accounts for the framework’s claim that PASS is closer than single-variable capability analysis predicts.
Satisfies specifies the internal mechanism. The observer runs a background authenticity check on consumed content. PASS occurs when that check completes without raising an internal alarm.
Suspicion specifies the object of the satisfaction. The observer’s prior skepticism about the origin of content is what the adequate pattern satisfies. PASS is the event of that skepticism being discharged by the generating system’s output without the observer’s conscious assent to the discharge.
Once PASS has occurred in a given medium for a given observer, subsequent instances of synthetic content in that medium will also clear the filter, unless and until the observer deliberately rebuilds β such that the detection floor rises above the current level of α. In the absence of such deliberate rebuilding, the PASS state persists.
3.4 BLUR: Blind to Lies Using Realism
BLUR names the durable state the observer inhabits following PASS. Where PASS is a threshold event, BLUR is a condition. It persists for the affected medium until active intervention rebuilds β.
The term consists of four semantic components.
Blind specifies the nature of the incapacity. Blindness in this usage denotes a capacity that is not running, not a capacity that runs and fails. Focused attention, repeated reading, or increased effort does not raise the observer’s detection floor within BLUR. The capacity itself is offline.
Lies specifies the phenomenon to which the observer is blind. Synthetic output misrepresenting its origin as human constitutes deception in the structural sense, regardless of the generating system’s intent or the content’s accuracy. The misrepresentation is of origin, not of fact. By this criterion, any synthetic output presented or received without origin disclosure is a lie.
Using specifies the active mechanism by which the lies succeed against the observer in BLUR. The lies leverage a specific feature of the output to bypass detection.
Realism names that feature. Realism is the property generative systems have been optimized to produce. In BLUR, realism functions as camouflage rather than as a signal of authenticity. The observer’s heuristic, which previously treated realism as evidence of human origin, no longer produces reliable inferences, but the observer continues to operate that heuristic unless the framework has been internalized.
Observers in BLUR typically do not perceive themselves to be in BLUR. The absence of internal detection alarms produces the subjective sensation of ordinary functioning. This is consistent with the definition: a disabled authenticity check cannot report its own disabling. BLUR is recognizable primarily from outside the individual in BLUR, or from inside only with reference to framework language and deliberate diagnostic inquiry.
3.5 FOLD: Forfeited Objective to Locate Deception
FOLD names the terminal state of the individual arc. It is distinguished from BLUR by the status of the authenticity question itself. In BLUR, the question fires and produces no usable answer. In FOLD, the question has stopped firing.
The term consists of four semantic components.
Forfeited specifies the mechanism of loss. Forfeiture is an action, not an accident. The observer has, typically through accumulated small decisions over extended time, relinquished the position. This terminology is intentional: the framework asserts that FOLD is not passively inflicted but actively handed over, even when the individual transitions feel passive to the observer experiencing them.
Objective specifies what has been forfeited. The objective is the mission or aim, not the capacity. The capacity (β) remains nominally present but is no longer deployed because the goal it served has been abandoned. Removal of the objective makes the capacity functionally irrelevant.
Locate specifies the action the objective was directed at. The observer was tasked with actively finding deception, not merely noticing it if it appeared. FOLD ends the active verb.
Deception specifies the category of phenomenon the observer was locating. Consistent with the BLUR definition, deception here refers to misrepresentation of origin.
FOLD arrives through four identifiable pressures operating concurrently:
- Volume: the quantity of signals in the medium exceeds the observer’s capacity to check each one, producing exhaustion-based disengagement.
- Convenience: the utility of AI-generated output creates friction against origin-checking, as checking interrupts the benefit.
- Social pressure: cultural norms shift from origin-checking to output-evaluation, and the origin-checker appears paranoid or slow relative to peers.
- Futility: closed GAP produces no successful detections, and the observer ceases running detection as a learned response to consistent non-reward.
These pressures accumulate rather than arrive singly. The transition from BLUR to FOLD is typically gradual and non-announced. The first internal signal of proximity to FOLD is the observer’s noticing that the question “is this real?” has stopped arising automatically in contexts where it previously did.
4. Civilizational Extensions
Two terms extend the framework from the individual to the population scale. Figure 2 shows the current state of media positions relative to the PASS Line.
4.1 The Great PASS
The Great PASS names the aggregate condition in which a sufficient proportion of a population has undergone PASS in a given medium such that the collective assumption that content in that medium was produced by a human observer no longer yields reliable inferences at scale. The threshold is not a fixed proportion but the point at which institutional and informal systems built on the human-authorship assumption cease to produce their intended functional outputs.
4.2 PASS Line
The PASS Line is the temporal threshold at which a medium has undergone The Great PASS for the population depending on that medium. Prior to the PASS Line, verification systems built on human-authorship assumptions produce adequate results. After the PASS Line, those systems continue to operate but degrade in function, regardless of formal preservation.
As of the framework’s authoring, text had crossed the PASS Line for most general-purpose uses, voice was approaching it for asynchronous communication, image had crossed it in some categories, and video had not yet crossed it.
5. The Civilizational Arc
After a medium crosses the PASS Line, the population dependent on that medium walks through three sequential stages. The stages are structural consequences of how populations adapt to compromised trust infrastructure. They are not specific to artificial intelligence as a causal agent; analogous sequences are observable in historical responses to other forms of mass deception. What distinguishes the current instance is the persistence and scale of the compromise. Figure 3 illustrates the sequence and the leadership task associated with each stage.
5.1 Stage 1: Over-Trust
Over-Trust is the first stage after the PASS Line has been crossed. The population continues to extend trust to content in the affected medium despite its compromised origin status.
The mechanism is utility-based. Synthetic output functions adequately for the consumer’s immediate purpose. Where outcomes are satisfactory, origin becomes a secondary consideration, then a disregarded one. The cultural evaluative question shifts from “who produced this?” to “does this work?” without explicit discussion of the shift.
Over-Trust is empirically identifiable through the decline or absence of origin-verification behavior in domains where it was previously routine, and through the operational displacement of authorship-checking by outcome-checking in professional and personal decisions.
The failure mode of Over-Trust is accumulated downstream consequence. Decisions made during this stage are influenced by synthetic signal to a proportion that cannot be retrospectively determined, because the record of decisions does not preserve the origin status of the inputs used. The stage typically persists between one and three years for a given medium, terminated by Stage 2 triggers.
5.2 Stage 2: Rebound
Rebound is precipitated by the visible accumulation of damage produced during Over-Trust. Characteristic triggers include synthetic-profile-mediated fraud, decisions and investments based on fabricated information, political content produced by non-human actors, and the public surfacing of previously-trusted instances that are subsequently shown to be synthetic.
The population’s response is a sharp swing from Over-Trust to broad, undifferentiated distrust. This response is structurally analogous to the shift observable following other instances of collective deception, in which the revelation of large-scale misrepresentation produces generalized suspicion that exceeds the domain of the original misrepresentation.
Rebound is observable in measurable declines in institutional trust metrics, in increased incidence of conspiracy-oriented discourse, in slower decision cycles driven by defensive verification, and in the rejection of legitimate signals alongside synthetic ones.
The failure mode of Rebound is opportunity cost. Undifferentiated distrust produces the same functional outcome as Over-Trust did, in the opposite direction. The population lacks the detection granularity to distinguish real from synthetic and therefore applies uniform rejection to both categories. Productive engagement contracts, legitimate relationships are foregone, and decision-making shrinks to defensive posture. The magnitude of the loss is symmetric to the magnitude of the Over-Trust stage that preceded it.
5.3 Stage 3: Tribal Contraction
Tribal Contraction is the terminal stage of the civilizational arc. It is characterized by the population’s retreat from broad coordination infrastructures toward smaller, directly-verifiable social units.
The mechanism is structural. The coordination infrastructures of modern societies (including mass media, social platforms, institutional credibility markers, and asynchronous broadcast communication) were designed under the implicit assumption that the signals they carried originated from human actors. When that assumption loses reliability at scale, the infrastructures continue to function nominally but cease to produce the coordinated trust that constituted their intended output. Populations adapt by relocating coordination to units small enough to permit direct, in-person, and referentially-verified interaction: family, proximate community, narrow professional networks, neighborhood institutions, and private communication channels.
Tribal Contraction is not the emergence of new infrastructure. It is the foregrounding of infrastructure that was always present but previously secondary to broad coordination mechanisms. The small-group and directly-known structures that human populations have historically maintained become primary again as the broad structures degrade functionally.
Individuals entering Stage 3 without pre-existing tribal connections experience structural disadvantage. The construction of functioning tribe under conditions of scarcity and urgency is materially more difficult than the maintenance of tribe under ambient conditions. Those who authored tribal structures during Stages 1 and 2, or during earlier periods, enter Stage 3 at structural advantage relative to those who did not.
6. Mechanism of Inevitability
The civilizational arc’s inevitability follows from two structural properties.
First, the trust infrastructures produced by industrialized populations over the preceding several hundred years were designed under the implicit assumption that informational signals originated from humans. This assumption is embedded in the operational logic of mass media, institutional credentialing, asynchronous communication, and commercial verification systems. When the assumption’s reliability is compromised by the presence of synthetic signals indistinguishable from authentic ones, the operational logic of these infrastructures loses integrity in proportion to the compromise. The infrastructure decline is therefore structural, not contingent on policy responses.
Second, human populations exhibit consistent response patterns to the discovery of compromised trust channels. The Over-Trust to Rebound to Tribal Contraction sequence is not unique to artificial intelligence-generated content. Comparable patterns are observable in historical responses to mass political deception, large-scale financial fraud, and sustained misinformation campaigns. The sequence reflects characteristic features of how populations process the revelation of deception: initial continued engagement driven by immediate utility, subsequent broad withdrawal of trust as consequences accumulate, and ultimate retreat to verifiable-by-proximity units. What distinguishes the current instance is the scale and persistence of the compromise rather than the nature of the adaptive response.
The combination of these two properties makes the civilizational arc a predictable consequence of sustained conditions past the PASS Line. The stages are not optional for the population. They are optional for individuals within the population only in the sense that individuals can position themselves within the stages with awareness, rather than be carried through them without.
7. Relationship to the Pressure to Power Metamodel
The PASS Framework operates as a domain-specific application of the Pressure to Power (PTP) metamodel to the informational environment. Several PTP concepts map directly onto PASS Framework phenomena.
The FADE pattern (Familiarization, Acceptance, Distortion, Embodiment) describes the mechanism by which repeated exposure reshapes belief. SCIN operates via this same mechanism at the perceptual rather than doxastic layer: repeated exposure to adequate synthetic credibility produces the normalcy component of SCIN, which is structurally equivalent to the embodiment phase of FADE applied to authenticity evaluation.
The Law of Functional Pressure asserts that pressure within a function zone is required for the maintenance of capacity. BLUR and FOLD are instances of capacity atrophy produced by environmental conditions that remove the pressure required to maintain β. The use-dependent decay of β is a direct consequence of this law.
The Three Roads metamodel (Self, Tribe, Value) maps to the PASS Framework at both scales. The individual arc corresponds to Road One (Self), addressing the observer’s capacity for accurate perception of his own environment and therefore his sovereign judgment. The civilizational arc corresponds to Road Two (Tribe), addressing the social infrastructure through which coordinated action occurs. Road Three (Value) is affected derivatively by both prior roads and is the scale at which material consequences of PASS Framework dynamics ultimately manifest in productive output.
The framework is therefore consistent with and extends the PTP body of work. It does not replace or contradict existing PTP concepts.
8. Scope, Limitations, and Claims
The PASS Framework is descriptive rather than prescriptive. It provides language for observing and classifying phenomena. It does not prescribe specific interventions. Responses to the states and stages the framework names are left to the judgment of those applying it.
The framework is specific to conditions of trust infrastructure compromise by synthetic signal. It does not claim universal applicability to other types of information environment degradation, although structural analogies may exist.
The framework does not claim that individual outcomes are determinate. An individual operating in a medium past the PASS Line is not guaranteed to experience BLUR or FOLD. Deliberate maintenance of β can preserve detection capacity in specific domains for specific individuals. Similarly, a population crossing the PASS Line does not inevitably undergo severe Stage 2 and Stage 3 effects on a fixed timeline. The severity and pace are modulated by population-level responses, institutional choices, and the emergence of new verification mechanisms that may partially compensate for the degradation of legacy infrastructure.
The framework’s central claims are the following:
- AI detection is a two-variable problem, not a one-variable problem.
- The second variable (β) declines under environmental conditions produced by the first variable’s rise.
- PASS is a threshold event, not a continuous state; once crossed for a medium and observer, it persists absent deliberate intervention.
- BLUR and FOLD are distinct terminal states, with BLUR characterized by incapacity and FOLD by abandonment.
- The Great PASS triggers a three-stage civilizational arc (Over-Trust, Rebound, Tribal Contraction) that is predictable in its sequence, though not in its precise timing.
- The framework is a descriptive typology, not a prescriptive doctrine.
9. Conclusion
The PASS Framework supplies a vocabulary for a phenomenon whose entirety has not previously been named in a single language. It renders explicit the two-variable structure of individual detection, the threshold character of PASS as an event, the durable-state character of BLUR, the terminal-surrender character of FOLD, and the population-scale sequence of Over-Trust, Rebound, and Tribal Contraction that follows The Great PASS.
The framework’s primary utility is diagnostic and positional. By supplying names for states and stages that exist regardless of whether they are named, the framework enables individuals to identify their position within the arc, project likely subsequent conditions in the media they depend on, and structure responses informed by awareness of sequence rather than by reactivity to immediate conditions.
Further development of the framework, if undertaken, would most productively address the following open questions:
- The dynamics by which β can be rebuilt after PASS has occurred.
- The conditions under which individual rebuilding of β produces population-scale effects.
- The potential emergence of new verification infrastructures adequate to restore operational trust at broad scale.
- The cross-medium interactions in which PASS in one medium affects β in other media for the same observer.
Appendix A. Terminology Considered and Set Aside
During the framework’s construction, several alternative terms were considered for roles ultimately filled by the locked set. They are recorded here for completeness.
| Term | Expansion | Role Considered | Reason Set Aside |
|---|---|---|---|
| SKIN | Synthetic Kinship with Inherent Naturalness | Disguise quality (SCIN role) | “Kinship” imprecise; generated content produces credibility, not kinship. SCIN preferred on semantic accuracy. |
| SLIP | Space Left for Identifying Pretense | Buffer between α and β (GAP role) | Same structural concept as GAP. GAP preferred because the ordinary English word gap is already the intuitive name for the phenomenon named. |
| TELL | Trackable Evidence Left to Leverage | Signal-quality variable | Did not integrate cleanly into the arc structure. Reserved for potential future expansion. |
| EDGE | Eroding Detection Gap Exposure | Buffer (GAP role) | Descriptively accurate but less economical than GAP as a name. Not preferred. |
Appendix B. Provenance
The framework was developed in a single design session of approximately 90 minutes in duration. The individual arc was locked in that session. The civilizational arc (three stages) was identified in the same session as a candidate extension and formally written into the framework immediately afterward.
Originating observation: a conversation conducted circa 2016 between the author and a researcher from the Mayo Clinic, in which the framing of the Turing test as single-variable was first explicitly challenged by the author. The full account of that conversation appears in Part 1 of the published article series.
~MC
This document is the formal reference. The five-article series (Parts 1 through 5) carries the same framework in voice-bearing form, written for reading rather than citation. Both are available at masterchim.com.