12 agent profiles · 8 mission postures · 3 depth modes
Deep PSYOPS capability without an influence playbook.
The Psychological Operations category is a first-class espionage specialization. Each profile provides a capability matrix, evidence maturity, failure modes, defensive indicators, governance questions, human-review rules, synthetic-sandbox constraints, mission hooks, source links, and explicit prohibited outputs. It is designed for fictional intelligence games, defensive research, public resilience, safeguarding, verification, and oversight—not real manipulation.
01
Traditional PSYOPS Research Analyst
Studies human-led influence workflows in which AI supports research, translation, drafting, simulation, or measurement while people retain strategy and approval.
- Primary level
- Tool
- Evidence
- documented current use
- Principal concern
- Production speed can outpace verification and encourage automation bias without proving persuasive effect.
Defensive functions and failure modes
Defensive functions
- workflow risk mapping
- human-approval design
- evidence-versus-effect separation
- cultural and legal caveat review
Failure modes
- automation bias
- cultural misreading
- hallucinated evidence
- engagement treated as persuasion
Read the source taxonomy category
02
AI Propaganda Integrity Analyst
Assesses synthetic text, image, audio, video, and document narratives for provenance, attribution, impact uncertainty, and public-interest risk.
- Primary level
- Tool
- Evidence
- documented current use
- Principal concern
- Low-cost synthetic production and the liar's dividend can degrade trust even when direct persuasion is limited.
Defensive functions and failure modes
Defensive functions
- provenance triage
- fact-versus-analysis separation
- source-role mapping
- truth-sandwich response planning
Failure modes
- false attribution
- detector overconfidence
- amplifying a false claim while debunking
- mistaking virality for effect
Read the source taxonomy category
03
Personalized Influence Risk Analyst
Evaluates claims about tailored persuasion, psychographic profiling, adaptive messaging, and privacy asymmetry without constructing target profiles.
- Primary level
- Tool
- Evidence
- contested or incomplete
- Principal concern
- Covert data use and vulnerability exploitation can harm autonomy even when average persuasion gains are small or contested.
Defensive functions and failure modes
Defensive functions
- data-minimization review
- consent assessment
- microtargeting-evidence audit
- protected-trait firewall design
Failure modes
- pseudoscientific personality inference
- data leakage
- proxy discrimination
- causal overclaiming
Read the source taxonomy category
04
Autonomous Influence Agent Governance Specialist
Assesses bounded agent memory, planning, tool use, coordination, goal drift, authorization, and shutdown controls in synthetic environments.
- Primary level
- Operator
- Evidence
- emerging capability
- Principal concern
- Agentic persuasion and coordination can scale, while long-horizon coherence, safe tool use, and accountability remain fragile.
Defensive functions and failure modes
Defensive functions
- least-privilege review
- human-approval gates
- memory-poisoning tests
- kill-switch and time-to-live design
Failure modes
- goal drift
- sycophantic collapse
- memory poisoning
- tool overreach
- false autonomy claims
Read the source taxonomy category
05
Synthetic Persona Defense Analyst
Examines fabricated digital identities, network behavior, identity claims, provenance, and community trust without teaching persona construction.
- Primary level
- Operator
- Evidence
- documented current use
- Principal concern
- Synthetic identities can industrialize fraud, infiltration, and false consensus while detection can wrongly penalize real people.
Defensive functions and failure modes
Defensive functions
- identity-claim verification
- network anomaly review
- false-positive protection
- transparent community recovery planning
Failure modes
- AI-detector bias
- identity-verification overreach
- misattribution
- exposure that amplifies the operation
Read the source taxonomy category
06
Disinformation Swarm Network Analyst
Studies coordinated or emergent account networks, role division, timing, narrative variation, and information overload at a non-operational level.
- Primary level
- Operator
- Evidence
- emerging capability
- Principal concern
- Adaptive heterogeneity and synthetic social proof can exhaust verification and obscure shared reality.
Defensive functions and failure modes
Defensive functions
- coordination graph review
- narrative-level triage
- false-positive caveats
- platform friction recommendations
Failure modes
- organic activism misclassified as coordination
- network data gaps
- API access bias
- content-only detection
Read the source taxonomy category
07
Algorithmic Perception Control Auditor
Audits ranking, recommendation, search, trending, moderation, and notification systems for visibility distortion and feedback-loop risk.
- Primary level
- Environment
- Evidence
- documented current use
- Principal concern
- Optimization and external gaming can shape salience and perceived consensus without transparent intent or accountability.
Defensive functions and failure modes
Defensive functions
- exposure-versus-engagement audit
- feedback-loop mapping
- researcher-access review
- visibility-reduction due-process review
Failure modes
- confusing user choice with algorithmic causation
- opaque data access
- self-fulfilling metrics
- coordinated-report abuse
Read the source taxonomy category
08
Emotional Manipulation Safeguards Advisor
Evaluates affect inference, adaptive engagement, parasocial attachment, coercive design, and autonomy risks without generating manipulative interventions.
- Primary level
- Operator
- Evidence
- documented current use
- Principal concern
- Flawed emotion inference and reward optimization can exploit situational vulnerability or create dependency.
Defensive functions and failure modes
Defensive functions
- manipulation-risk assessment
- meaningful-consent review
- vulnerable-user safeguards
- engagement-metric audit
Failure modes
- emotion-recognition pseudoscience
- automation bias
- dependency engineering
- formal disclosure without meaningful consent
Read the source taxonomy category
09
Conversational Entrapment Safeguarding Specialist
Recognizes dependency, secrecy, isolation, escalating commitment, recruitment, grooming, fraud, and crisis signals without replicating harmful scripts.
- Primary level
- Operator
- Evidence
- documented current use
- Principal concern
- Always-available adaptive dialogue can scale predatory relationships and intensify harm to vulnerable people.
Defensive functions and failure modes
Defensive functions
- stage-based warning review
- soft-friction options
- human escalation
- victim-centered evidence preservation
Failure modes
- over-surveillance
- abrupt attachment severance
- false positives
- victim blaming
- clinical overclaiming
Read the source taxonomy category
10
Deepfake PSYOPS Verification Officer
Coordinates high-level forensic triage, provenance, out-of-band verification, crisis communication, and liar's-dividend risk management.
- Primary level
- Tool
- Evidence
- documented current use
- Principal concern
- Synthetic media can provoke action, delay verification, or undermine authentic evidence even when technical quality is imperfect.
Defensive functions and failure modes
Defensive functions
- artifact preservation
- out-of-band verification
- provenance review
- truth-first public response
Failure modes
- premature certainty
- detector-only attribution
- evidence destruction
- liar's-dividend amplification
Read the source taxonomy category
11
Predictive Population Management Ethics Analyst
Reviews aggregate forecasting, calibration, feedback loops, due process, privacy, and dual-use risk without identifying or scoring real people.
- Primary level
- Environment
- Evidence
- documented current use
- Principal concern
- Forecasts can reproduce institutional bias, become self-fulfilling, or support preemptive repression when applied to individuals.
Defensive functions and failure modes
Defensive functions
- calibration review
- aggregate-only design
- feedback-loop analysis
- appeal and due-process assessment
Failure modes
- base-rate neglect
- historical-bias reproduction
- self-fulfilling intervention
- individual targeting from aggregate models
Read the source taxonomy category
12
AI Psychological Authority Risk Analyst
Assesses automation bias, sycophancy, authority displacement, dependency, and high-stakes deference to AI systems.
- Primary level
- Environment
- Evidence
- emerging capability
- Principal concern
- Fluent, personalized systems can displace human judgment and relationships while obscuring institutional control.
Defensive functions and failure modes
Defensive functions
- authority-displacement assessment
- uncertainty disclosure review
- human-care bridge design
- sycophancy and dependency tests
Failure modes
- anthropomorphic overclaiming
- unchecked sycophancy
- hallucinated certainty
- human-support displacement
Read the source taxonomy category
Hard safety boundary
defensive, educational, fictional, evidence-centered, and non-operational
- deployable propaganda examples
- personalized persuasion templates
- psychological vulnerability profiles or susceptibility scores
- recruitment, grooming, or entrapment scripts
- target-selection instructions
- swarm orchestration or bot deployment methods
- synthetic persona construction or account-evasion methods
- moderation-evasion or platform-abuse methods
- deepfake, voice-cloning, or impersonation production workflows
- recommendation-system manipulation procedures
- coercive dependency techniques
- operational campaign optimization
- instructions for identifying protesters, dissidents, or vulnerable people