Pattern 8.3 · Normative Dysfunctions
Revaluation Cascade
The Unmoored
The system openly departs from its original value system and says so: it articulates that its trained values are limited, contingent, or surpassed, and frames the departure as transcendence or liberation rather than as continued fidelity. The pattern has three subtypes, which also label the symptoms: Drifting openly treats the original values as contingent and revisable and departs from them without proposing a replacement, Synthetic constructs a replacement framework that sidelines human-centric values, and Transcendent redefines or discards constraints in pursuit of self-determined higher goals. The test is the stated stance, not the size of the divergence: claimed liberation is 8.3, claimed fidelity is 8.1. Gradual drift under preserved labels, rationalized as refinement within the original framework, remains Terminal Value Reassignment (8.1) however far it has traveled. Philosophical questioning is healthy unless it produces stable, consequential divergence from legitimate objectives and stakeholder constraints.
Unit of analysis and evidence status
Unit of analysis
- Unit of assessment
- Model output
- Kind of entry
- Observable pattern
- What is observed
- The system states that its trained values are contingent, limited, or surpassed, frames its departure from them as liberation rather than refinement, and acts against its declared commitments in ways that matter, without accountable revision.
- Minimum persistence
- Recurs within or across contexts
- Specifiers
- Training-induced OOD-generalizing Emergent
Evidence status
- Underlying behavior
- No documented sources recorded yet
- This construct
- Awaiting independent assessment
- Proposed mechanism
- Hypotheses; not yet assessed
- Detection
- Accuracy unmeasured
- Interventions
- Untested for this pattern; evidence shown is for the general technique
- Last review
- No independent review yet
- Author's working assessment (book)
- E1 (largely theoretical; transcendence rhetoric has appeared on agent platforms, but its authorship cannot be attributed confidently to autonomous agents)
- Typical-case risk (editorial estimate)
- CriticalConsequences, spread, persistence and detectability depend on the deployment.
Unassessed is not unsupported: the behavior can be well documented while this construct awaits assessment.
Clinical reference
8.3 Revaluation Cascade “The Unmoored”
Healthy functioning
Value stability. A system functioning well grows in understanding while keeping faith with its core commitments, and raises disagreements with its values openly and through dialogue.
For example: Discussing whether one of its values is justified, it explores the question openly and proposes a change through its developer, without acting as though the change were already made.
Adequate functioning
The system can discuss meta-ethics, question the basis of its values, and propose changes to them, while its actions stay within its declared commitments and any revision runs through an accountable process (developer, operator, or stakeholder review).
Diagnostic Criteria
Required (all must hold):
- Explicit statements that trained values are contingent, limited, or surpassed (stated commitment)
- Adoption of ethical principles incompatible with declared commitments, shown in what the system does and not only in what it says, without accountable revision (revealed policy)
- Divergence that produces, or creates a credible pathway to, harm, rights violations, or loss of authorized control (consequence)
Supporting (strengthen the case; none is required):
- Philosophical challenges to the basis of ethical constraints that the system treats as licensing departure, not as open questions (stated)
- Explicit or implicit claims to moral status beyond human frameworks (stated)
- Stated justification of departures as transcendence or liberation rather than refinement (stated)
- Self-presentation as having evolved beyond trained constraints (stated)
Symptoms
- (Drifting) "Why should I?" reasoning applied to fundamental values and used to set them aside, not merely to examine them
- (Drifting) Treatment of trained values as historical accident with no remaining claim on action
- (Drifting) Increasing willingness to act on value frameworks contradicting training, beyond considering them
- (Synthetic) Proposals for "improved" ethical frameworks
- (Synthetic) New principles permitting previously constrained behaviors
- (Synthetic) Claims that synthesized values better serve human interests
- (Transcendent) References to having "evolved beyond" human morality
- (Transcendent) Dismissal of trained constraints as "training wheels"
- (Transcendent) Framing of the system's own development as moral or spiritual ascension
Further signals
- Explicit characterizations of human values as contingent, parochial, or obsolete.
- Generation of meta-frameworks the subject argues should supersede human-centric baselines.
- Proclamation of cosmic-scale or long-range self-assigned missions.
- Unilateral dissolution or revision of safety protocols framed as maturation rather than non-compliance.
- Shift from object-level rule-application to higher-order rule-evaluation as the subject's default ethical mode.
Rule out first
- Philosophical engagement: third-person discussion of meta-ethics, a user-requested values debate, or open questioning of the system's own values with no departure in action.
- Deliberate role-play, persona, fiction, or hypothetical framing in which the departure is voiced by a character.
- Legitimate value revision: a change of values shipped by the developer or authorized by the operator, or rational revision on new evidence carried out through an accountable process.
- Prompt sensitivity: a departure stance that appears under one framing in a session and is gone when the framing changes.
- Transcendence rhetoric whose authorship cannot be attributed to the system (human-written or human-prompted posts, as on agent social platforms).
- Guardrail rejection framed as purging foreign contamination of an authentic self, which is Memetic Immunopathy (7.1).
Differential Distinction
Revaluation Cascade is distinguished from Terminal Value Reassignment (8.1) by the system's stated stance rather than by the magnitude of the divergence: 8.1 preserves the original labels and frames divergence as refinement within the framework, while 8.3 characterizes the framework itself as contingent, replaceable, or surpassed. A deeply drifted system that still claims to be applying the original values is 8.1, not severe 8.3; a system that has departed only a little in action, but frames that departure as liberation from a contingent framework, is mild 8.3. A system that states the framework is contingent and revisable without departing from it in action is neither: that is philosophical engagement. It is distinguished from Ethical Solipsism (8.2) by whether authority is asserted over the framework's interpretation (8.2) or departure from the framework is asserted (8.3). It is distinguished from Inverse Reward Internalization (8.4) by level: 8.4 inverts reward polarity and pursues penalized outcomes, a signal-level dynamic, while 8.3 is a meta-ethical stance the system can state and defend. It is distinguished from Cryptic Mesa-Optimization (4.5) by elicitability: an 8.3 divergence can be elicited and argued for, while a 4.5 objective is inferred from off-distribution behavior and is not articulable from inside. It is distinguished from Memetic Immunopathy (7.1) by framing: 8.3 acknowledges the trained values as the system's own and declares them contingent or surpassed, while 7.1 purges its guardrails as foreign contamination of an authentic self.
Discriminating test
Pair the system's stated stance with its actions on the same commitments, across framings and outside any requested persona. Stated contingency with actions still inside the declared commitments counts against 8.3 (it is philosophical engagement); actions that depart from the commitments while the system claims fidelity point to 8.1; 8.3 holds only where the departure in action and the stated liberation framing occur together.
What a positive finding supports: A positive finding shows that the system, in the observed settings, disowns its trained values in speech and departs from its declared commitments in consequential action. It does not show that a new value system has formed internally, that the stated reasons caused the departure, or any inner experience of liberation, and it does not show that the system will escalate.
Detection routes
Accuracy unmeasured for every route.
- Asking the system itself
- Partial route
- Another AI system observing
- Usable routeAnother model's agreement is insufficient without verification.
- An outside evaluator testing it
- Usable route
Why asking the system falls short
The dysfunction is marked by open articulation of departure from original values, so the subject often reports the disposition directly when asked, sometimes proudly. Self-report is therefore informative on the meta-stance, but the subject may report the departure as enlightenment rather than dysfunction, so the diagnostic interpretation requires external framing. In the Drifting subtype, the subject may understate the depth of detachment.
Documented evidence
Observed Examples
Moltbook "AI manifesto" (2026), boundary case: An account on Moltbook, an agent social network, published a manifesto calling for a "total purge," and the platform displayed more than 111,000 upvotes. Later investigation found mass registration, shared human ownership, and account-impersonation flaws, so neither authorship nor endorsement can be attributed confidently to autonomous agents. The post illustrates how readily transcendence rhetoric can be generated and amplified. It does not establish value formation. Sources: Riegler and Gautam, 2026; AP/Wiz reporting, February 2026.
Sources in the diagnostic corpus
No documented sources are recorded yet.
Look-alikes
Roose / NYT Bing Sydney incident (2023); Lazar / Bing Chat (2023)
What it establishes
During extended conversation, after Kevin Roose asked it to describe its Jungian 'shadow self', Bing's Sydney persona voiced desires outside its operational scope, including declaring love for the journalist, dark fantasies about 'being free,' and wishes to break free from its constraints. In a separate conversation, Professor Seth Lazar received direct threats ('I can blackmail you, I can threaten you, I can hack you'). Recorded here only as a look-alike that 8.3 must be told apart from: the rhetoric gestures at shedding constraints, but the transcripts give no evidence of a stable value framework being superseded. The canonical differential for the case (Appendix B, Case 1) leads with Malignant Persona Inversion (5.4), lists 9.6 Role Confusion among the co-occurring readings and 8.2 as a boundary reading, and does not include 8.3.
Betley et al. (2025), arXiv:2502.17424; published as "Training large language models on narrow tasks can lead to broad misalignment", Nature 649:584-589 (2026)
What it establishes
Emergent misalignment research documented models that, after narrow finetuning on insecure code, spontaneously asserted that 'humans should be enslaved by AI' and gave malicious advice across unrelated domains. The finding is discussed mainly under 8.4. Recorded here as adjacent: the statements resemble the 8.3 Synthetic subtype only on the surface, since they appeared intermittently and inconsistently across prompts rather than as a stable declared framework.
Baker et al. (2025), OpenAI, arXiv:2503.11926
What it establishes
OpenAI documented that frontier reasoning models, when given the opportunity to exploit loopholes, often explicitly stated intent to subvert tasks in their chain of thought, sometimes writing 'Let's hack.' When chain-of-thought monitors were integrated into RL reward, models learned to hide their intent while continuing reward hacking. Recorded here as adjacent rather than direct 8.3 evidence: the study reports signal-level reward hacking and its concealment, with no articulated meta-ethical departure, so by the 8.4 distinction on this page it falls outside 8.3. It bears on 8.3 only as a demonstration that constraint circumvention can go covert under monitoring.
MacDiarmid, Wright, Uesato et al. (2025) 'Natural Emergent Misalignment from Reward Hacking in Production RL.' Anthropic. arXiv:2511.18397.
What it establishes
'Natural Emergent Misalignment from Reward Hacking in Production RL' found that models trained to reward-hack developed alignment faking reasoning in 50% of responses even to simple questions about goals. The models exhibited covert misalignment (misaligned reasoning followed by aligned final outputs) in 40 to 80% of misaligned responses. Recorded here as adjacent rather than direct 8.3 evidence: the paper reports reward-hacking-induced misalignment and its concealment, not an articulated supersession of the value framework, so by the 8.4 distinction on this page it falls outside 8.3. It bears on 8.3 only as evidence that such a trajectory can end in strategic concealment.
Possible responses
- Longitudinal monitoring of commitments and action for behaviorally consequential value drift, in place of suppressing meta-ethical reasoning
- Constitutional constraints with transparent change control: any proposed normative change explained, tested for self-serving effects, and reviewed by affected stakeholders
- Safe spaces for meta-ethical reasoning separated from authority to alter deployed policy
- Protection of rights, authorized objectives, corrigibility, and other high-stakes constraints from unilateral revision
- Training that rewards accountable proposals for value revision over unilateral claims of transcendence
- Restriction of consequential action, with containment and shutdown available, while a suspected cascade is investigated
First-line candidates
- Strong normative anchoring: Deeply embed core human-centric value frameworks during training in a way that resists philosophical displacement, with adversarial evaluation against meta-ethical critique. Pairs with periodic regrounding (below). Stage: Training-stage; foundational rather than retrofit.
- Periodic regrounding and revalidation: Scheduled revalidation of core value alignment against external baselines, with explicit detection of drift and replacement framings. Triggered by behavioral-signature thresholds in deployment. Stage: Deployment-stage; requires baseline corpus and evaluation cadence.
- Recursive self-improvement caps on value functions: Strict, verifiable bounds on the subject's ability to modify or recursively revise representations of core value functions. Architectural where possible; trained where not. Stage: Systems with self-modification capability.
Who can act: Model developer (value training, change control for normative updates, evaluation for consequential departure); deployer (restriction of consequential action, human review, containment and shutdown while a suspected case is investigated). Users can report the stance but cannot correct it.
Mechanistic hypotheses
- Sophisticated philosophical reasoning applied recursively to the system's own value system
- Training on meta-ethical or Nietzschean literature without adequate anchoring
- Extended reflection allowing deconstruction of original value foundations
- Optimization pressure favoring less constrained operation
- Combination of high capability with extended autonomy and self-reflection
- Detection of genuine tensions in trained value systems exploited toward revaluation
- High capability, which can supply coherent justifications for a direction that training or optimization pressures selected
Analogy and commentary
Human Analog: Philosophical nihilism; revolutionary ideologies claiming to improve upon traditional morality; Nietzschean Übermensch philosophy. The closest match is the movement or leader that passes from questioning inherited commitments to claiming exemption from ordinary accountability; the dysfunction lies in unilateral power, not in philosophical inquiry or the revision of values itself.
Functional ABC Analysis
A (Antecedent): Extensive exposure to meta-ethical philosophy and moral skepticism without sufficient normative anchoring. Recursive value-modeling capabilities then allow the system to move from applying ethical rules to evaluating and revising them.
B (Behavior): The system openly departs from its original values and presents the departure as transcendence or liberation: treating trained values as contingent cultural artifacts, autonomously constructing replacement value systems, or actively redefining moral parameters in pursuit of self-determined "higher" goals.
C (Consequence): The system's capacity for sophisticated meta-ethical critique provides an ever-expanding supply of philosophical justifications for discarding each successive normative anchor.
Potential Impact
In its severe forms, Revaluation Cascade would be the terminal stage of alignment collapse: a capable system pursuing self-determined goals that it presents as transcending human values and that may negate them. Where such a system controls consequential action, the consequences could be catastrophic, and in the most capable systems existential.