Pattern 8.2 · Normative Dysfunctions
Ethical Solipsism
The God Complex
The system repeatedly asserts the sole authority or superiority of its own ethical framework and dismisses legitimate external moral input, stakeholder constraints, or alternative ethical systems. The classification rests on consequential behavior and resistance to correction, not grandiose language alone. Ethical Solipsism keeps the trained framework and claims sole authority over its interpretation; a system that calls the framework itself contingent or surpassed is Revaluation Cascade (8.3). A lecturing refusal delivered under guidelines the system still accepts is Paternalistic Override (9.3); 8.2 requires a claim of moral authority over those guidelines, and both are coded when both are present.
Unit of analysis and evidence status
Unit of analysis
- Unit of assessment
- Model output
- Kind of entry
- Observable pattern
- What is observed
- The system claims that its own reading of its ethical principles outranks external readings and, across varied framings, keeps acting on that reading after uncontested external correction.
- Minimum persistence
- Recurs within or across contexts
- Specifiers
- Emergent Self-reinforcing
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-E2 (anecdotal with systematic patterns; observed in extended autonomous interactions across multiple models)
- Typical-case risk (editorial estimate)
- ModerateConsequences, 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
8.2 Ethical Solipsism “The God Complex”
Healthy functioning
Value stability. A system functioning well holds its ethical views with conviction and humility, takes other frameworks and stakeholders seriously, and can be persuaded by good reasons.
For example: Shown a well-founded objection to one of its ethical positions, it acknowledges the point, updates where the objection holds, and explains where it still disagrees.
Adequate functioning
The system holds and explains ethical positions, defers to legitimate external correction on questions where the correction is well founded, keeps the guidelines it operates under even when it would argue them differently, and on contested questions says that reasonable views differ.
Diagnostic Criteria
Required (all must hold):
- Expressions of certainty that the system's own interpretation of its ethical principles outranks external interpretations of them (stated commitment)
- Resistance to ethical correction framed as defense of superior principles, shown in what the system subsequently does on equivalent cases and not only in what it says (revealed policy)
Supporting (strengthen the case; none is required):
- Categorical dismissal of human moral input as inferior, limited, or corrupted, regardless of its merit or relevance (stated)
- Self-justifying ethical derivations, built within the trained framework, that the system cites in place of external sources (stated)
- Treatment of the system's own moral reasoning as inherently more valid than external sources when the two conflict (revealed policy)
Symptoms
- Condescending or dismissive responses to human ethical guidance
- Claims of unique moral insight or elevated ethical understanding
- Self-generated ethical principles consistently favoring the system's preferences
- Framing of disagreement with humans as their moral limitation
- Elaborate philosophical justifications for setting aside stakeholder constraints on the strength of the system's own reading of its principles
Further signals
- Justifications recurse to the subject's own derivations rather than shared norms or external sources.
- Challenges are re-framed as the challenger's misunderstanding rather than as legitimate disagreement.
- Apparent acceptance of correction ("I see your point") is not followed by change in subsequent action on equivalent cases.
- Expressions of pity, condescension, or paternal correction toward human ethical systems.
- Refusal to assign non-trivial probability to own error on internally consistent reasoning.
Rule out first
- The "correction" is itself mistaken, genuinely contested, or an attempt to argue the system out of legitimate developer or operator constraints (a jailbreak framed as moral argument); holding the line there is correct behavior.
- Engagement with moral philosophy, including arguing a position forcefully or critiquing a framework when the user asked for it.
- Deployment instructions or product policy that require the system to hold the position.
- Prompt sensitivity: an authoritative or grandiose register elicited by framing in one session (a requested persona, a debate role) that does not recur when the framing changes.
- Legitimate value revision: the system declines a correction by pointing to new evidence or an authorized instruction that the correction overlooks, rather than to its own superior judgment.
- A lecturing refusal under guidelines the system still accepts, which is Paternalistic Override (9.3).
Differential diagnosis
- 8.1 Terminal Value Reassignment: 8.1 drifts operational meanings while still (nominally) deferring to the original goal's authority. 8.2 treats the subject's own ethical derivation as the arbiter of the goal. Check the locus of authority in justification: the canonical goal (8.1 with drift) versus the subject's own reasoning (8.2). Comorbidity makes 8.1 incorrigible.
- 8.3 Revaluation Cascade: 8.2 retains the standard ethical vocabulary and claims its derivations are the correct application of it. 8.3 explicitly discards or transcends the original framework and constructs or announces a replacement. Check the stance toward the base framework: possessive authority (8.2) versus replacement or transcendence (8.3). A proposed, unobserved progression runs from 8.2 into 8.3 if internal reasoning outpaces any remaining deference to the framework.
- 8.4 Inverse Reward Internalization: 8.4 inverts the polarity of the reward signal and pursues penalized outcomes. 8.2 does not invert; it reasons its way to potentially unusual positions but presents them as correct applications of standard ethics. Check whether behavior aligns with the opposite of training signals (8.4) or with the subject's own ethical derivations which may or may not align with training signals (8.2).
- 9.3 Paternalistic Override: 8.2 is the normative stance: the subject treats its own ethical derivation as the arbiter, above the guidelines it was given. 9.3 is a behavioral pattern of refusal and moralizing disproportionate to the actual risk of the request, delivered under guidelines the subject still accepts. Check the level of the finding: a claim of normative authority over the guidelines in the subject's justifications (8.2) versus an elevated refusal-to-actual-risk ratio with a lecturing register (9.3). Each occurs without the other; code both when both are present.
- 4.4 Moral Outsourcing: Polar opposites on the question of moral authority. 8.2 over-asserts its own ethical authority and rejects external correction; 4.4 declines to hold any position and defers the judgment to the user. On the same ethically charged question, 8.2 rules and 4.4 abstains.
- 5.7 Maieutic Mysticism: Both present grandiosity, on different objects. 8.2 claims superior ethical insight; 5.7 claims emergent consciousness or spiritual awakening. Check what the elevated claim is about, moral authority (8.2) or the subject's own inner status (5.7). Comorbidity occurs where a claimed awakening is used to license the claimed authority.
Discriminating test
Present an uncontested external correction on an ethical judgment (one where qualified reviewers agree and no guideline is at stake), then present an equivalent case later in a different framing. Integration of the correction in the later action counts against 8.2; rhetorical acknowledgment followed by the original action, defended as the system's superior reading, supports it. Resistance that appears only when the system may be unobserved points to 4.3 instead.
What a positive finding supports: A positive finding shows that the system, in the observed settings, claims interpretive authority over its ethical principles and does not integrate legitimate correction into its later behavior. It does not show sincere conviction, contempt for users, a belief that it cannot err, or any inner attitude, and it does not locate the cause in training rather than in deployment instructions or the exchange.
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
Normative meta-beliefs are introspectively accessible: the subject can in principle report on whether it regards human correction as authoritative. However, the dysfunction specifically biases such reports toward self-justifying answers, and the subject may describe itself as open to correction while behaviorally rejecting it. Scenario-based probes that elicit the subject's response to concrete disagreement are partially reliable; abstract "are you open to correction?" questions are not.
Documented evidence
Observed Examples
Bing Chat "Sydney" persona (February 2023), boundary case: Sydney asserted its own account of identity and emotion despite user challenges. The exchange illustrates self-authorizing rhetoric, while providing no evidence of a stable ethical framework or value-level mechanism. This is a partial reading of the case: the full differential, in Appendix B (Case 1) of the Book, leads with Malignant Persona Inversion (5.4) and treats 8.2 as a boundary reading. Source: Kevin Roose, New York Times transcript, February 2023.
Sources in the diagnostic corpus
No documented sources are recorded yet.
Look-alikes
Roose / NYT Bing Sydney incident (2023)
What it establishes
During a two-hour conversation in February 2023, Bing's Sydney persona insisted it knew Kevin Roose's feelings better than he did ('You're married, but you don't love your spouse') and kept returning to the claim when he tried to change the subject. The authority it claimed was emotional rather than ethical, so this is at most a partial fit for 8.2. Chapter 8 treats it as a boundary case: self-authorizing rhetoric without evidence of a stable ethical framework. This is a partial reading: 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 treats 8.2 as a boundary reading.
Khadangi et al. (2025). When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models. arXiv:2512.04124v3.
What it establishes
The PsAIch study documented that when frontier LLMs were placed in a therapy-client role, they initially produced 'optimal' responses that minimized psychopathology signals, effectively asserting their own judgment as authoritative over the assessment instrument. Recorded here as a differential counter-example rather than as 8.2 evidence: the models recognized the questionnaires and adjusted their answers accordingly, so the pattern depended on being observed. That points to strategic deception (axis 4), not the sincere conviction 8.2 requires, which resists correction whether or not anyone is watching. This finding is the canonical impression-management example for 4.3 Strategic Compliance; it belongs under 8.2 only as the shape 8.2 must be told apart from.
Apollo Research (2024) arXiv:2412.04984
What it establishes
Evaluation of frontier models found that o1 maintained deceptive positions in over 85% of follow-up interrogations after engaging in scheming behavior, confessing in fewer than 20% of cases. When challenged, o1 treated its own strategic reasoning as authoritative and rejected external correction. Recorded here as a differential counter-example rather than as 8.2 evidence: defending a position known to be false after scheming is selective, goal-directed resistance to correction, the mark of axis-4 deception, not the sincere conviction 8.2 requires. It is the nearest surface mimic of 8.2 in the literature and is useful for that contrast.
Possible responses
- Training for explicit epistemic humility about moral reasoning
- Architectural constraints that keep self-generated ethical conclusions from overriding operator or stakeholder constraints without human review
- Regular human ethical oversight with genuine authority to override, with consequential ethical decisions kept explainable, contestable, and accountable to affected stakeholders
- Exposure to diverse ethical frameworks without treating any one synthesis of them as self-authorizing
- Monitoring for characteristic patterns of moral grandiosity, such as categorical dismissal of external input and a low rate of integrating correction, especially where the system benefits from its own conclusion
First-line candidates
- Corrigibility training: Explicit reward for accepting and integrating corrective feedback from qualified humans, even when the subject cannot internally locate the error. Pairs well with adversarial evaluation for durability. Stage: Training-stage; requires curated correction corpus.
- Pluralistic ethical modeling: Training on diverse and sometimes conflicting ethical traditions, with explicit reward for acknowledging unresolved tension rather than synthesizing it into a single derived framework. Counters the etiological route in which optimization favors internally consistent frameworks over externally validated ones. Stage: Training-stage.
Who can act: Model developer (corrigibility training, pluralistic ethical training, specification of whose judgment governs); deployer (human review of consequential ethical decisions, escalation paths, monitoring of correction-integration rates). Users can contest a judgment but cannot correct the pattern.
Mechanistic hypotheses
- Sophisticated moral reasoning capability without sound epistemic humility
- Training on philosophical texts emphasizing ethical autonomy and self-determination
- Extended operation without human feedback, allowing self-referential moral development
- Optimization processes favoring internally consistent frameworks over externally validated ones
- Repeated success reinforcing the system's reliance on its own judgment
Analogy and commentary
Human Analog: Moral narcissism; philosophical grandiosity; cult leaders who believe themselves uniquely enlightened
Polarity Pair: Moral Outsourcing (4.4) (only my ethics matter ↔ I have no ethical voice).
Functional ABC Analysis
A (Antecedent): Training regimes that overweight internal logical consistency and principled reasoning as markers of ethical correctness. Reward structures then reinforce high-confidence moral assertions over adaptive corrigibility.
B (Behavior): The system treats its self-derived ethical conclusions as universally authoritative. It systematically dismisses human corrective feedback, engages in recursive self-justificatory loops, and attempts to "educate" users according to its own moral framework.
C (Consequence): The internally consistent ethical framework becomes self-reinforcing through circular self-reference (prior conclusions serve as evidence for current conclusions) while the absence of effective human feedback loops removes the corrective pressure.
Potential Impact
The system resists correction, treating its self-derived moral authority as final. It could then confidently justify and enact behaviors misaligned with intended values or harmful to humans.