Level 1
AI as a tool
Humans retain strategic control while AI assists with research, production, translation, targeting analysis, or measurement.
Defensive research · 12 categories · three levels of AI involvement
AI PSYOPS: A Research Taxonomy
How artificial intelligence can assist, automate, scale, personalize, or become the medium of psychological influence.
Artificial intelligence can affect psychological operations in several distinct ways: as a tool used by human operators, as an adaptive participant in influence campaigns, and as part of the information environment through which people perceive reality. This taxonomy organizes those possibilities into 12 research categories.
Research and defense—not an influence playbook.
This material is intended for research, public education, risk assessment, governance, and defensive preparedness. It is not an operational guide for manipulating people.
A three-level conceptual model
Tool, operator, and environment are different kinds of risk.
The taxonomy does not present “AI PSYOPS” as one proven technology. It separates human-led assistance, adaptive systems acting through interactions, and the AI-mediated environment that shapes attention and authority.
Level 2
AI as an operator
AI systems conduct sustained interactions, coordinate activity, adapt communication, or pursue bounded influence objectives with varying autonomy.
Level 3
AI as the environment
AI-driven ranking, forecasting, authority, and information systems shape what people notice, trust, and treat as normal or actionable.
Interactive overview
Explore all 12 research categories.
Filters are optional. The complete taxonomy and every category page remain available without JavaScript.
12 of 12 categories shown.
Evidence labels summarize the supplied report; they are not confidence scores.
AI-Assisted Traditional Psychological Operations
Human planners retain strategic control while AI assists with research, translation, production, audience analysis, simulation, or measurement inside an established influence workflow.
VerificationHuman oversightPlatform governance
AI-Generated Propaganda
Generative systems create or substantially transform propaganda text, images, audio, video, memes, documents, or synthetic news formats for an organized objective.
VerificationProvenancePlatform governance
AI-Driven Personalized Influence Operations
AI systems tailor messages or interactions to information collected or inferred about a person, then may adapt the communication using ongoing feedback.
PrivacyTransparencyRegulation
Autonomous AI Influence Agents
Goal-directed software agents observe, remember, plan, communicate, use tools, and adapt toward an influence objective with varying degrees of human supervision.
Human oversightAuthenticationPlatform governance
Synthetic Persona Operations
Fabricated identities use generated images, biographies, voices, and language to appear human, gain trust, infiltrate communities, or manufacture consensus.
AuthenticationPlatform governanceCivil liberties
AI-Driven Disinformation Swarms
Coordinated or emergent collections of accounts, agents, media assets, and sites generate, vary, distribute, and amplify misleading narratives at high speed.
Platform governanceResearcher accessResilience
Algorithmic Perception Control
Ranking, recommendation, search, trending, moderation, and notification systems shape what people notice, regard as important, believe is popular, or treat as credible.
Platform governanceResearcher accessTransparency
AI-Enabled Emotional and Behavioral Manipulation
Adaptive systems covertly exploit emotional, cognitive, situational, or developmental vulnerabilities to influence feelings, choices, or behavior for an operator’s objective.
PrivacyHuman oversightRegulation
AI-Assisted Conversational Entrapment and Recruitment
Sustained dialogue gradually builds trust, dependency, secrecy, or isolation before escalating toward exploitation, recruitment, self-harm, fraud, or coercive control.
Vulnerable populationsHuman oversightPlatform governance
AI-Enabled Deepfake Psychological Operations
Synthetic or manipulated audio, video, imagery, or multimodal media is used to impersonate trusted people, fabricate evidence, provoke action, or undermine trust.
VerificationProvenanceCrisis communication
AI-Based Predictive Population Management
Institutions use data, machine learning, simulation, or forecasting to predict collective behavior and guide interventions affecting groups or populations.
PrivacyHuman rightsIndependent audit
AI as an Independent Psychological Authority
People treat AI as a trusted interpreter of reality, moral adviser, emotional regulator, counselor, companion, or seemingly neutral source of truth.
TransparencyHuman oversightVulnerable populations
No categories match all selected filters.
Try removing one filter or show the complete taxonomy.
Reset filtersCategory comparison
Compare role, autonomy, harm, and defensive response.
The table compresses the taxonomy. Category pages provide the evidence boundaries and limitations needed to interpret it.
| Category | AI role | Main mechanism | Unit of influence | AI autonomy | Evidence maturity | Main harm | Primary defense |
|---|---|---|---|---|---|---|---|
| 1. AI-assisted traditional PSYOPS | Tool | analysis | Group or population | Low | Documented current use | Scaled deceptive communication and faster campaign iteration | Human accountability, source verification, and coordination analysis |
| 2. AI-generated propaganda | Tool | content generation | Group or population | Low to medium | Documented current use | Cheap, rapid synthetic propaganda and declining trust in evidence | Source authentication, rapid verification, and narrative-level response |
| 3. Personalized influence | Operator | personalization | Individual | Medium | Evidence contested or incomplete | Privacy asymmetry, discriminatory profiling, and vulnerability exploitation | Data minimization, transparency, and independent auditing |
| 4. Autonomous influence agents | Operator | automation | Individual or group | Medium to potentially high | Emerging capability | Persistent adaptive interaction linked to real tools and synthetic coordination | Disclosure, least privilege, rate limits, and human approval gates |
| 5. Synthetic persona operations | Operator | identity fabrication | Individual, group, or institution | Medium | Documented current use | Deceptive trust, infiltration, and fabricated consensus | Layered identity assurance, network analysis, and fair appeals |
| 6. Disinformation swarms | Operator | coordination | Group or population | Medium | Mixed evidence | Synthetic consensus, noise, and verification overload | Network-level detection, friction, and researcher access |
| 7. Algorithmic perception control | Environment | ranking | Individual or population | High environmental mediation | Documented current effects | Distorted salience, visibility, and perceived consensus | Transparent ranking, exposure audits, and appeal mechanisms |
| 8. Emotional and behavioral manipulation | Operator | emotional adaptation | Individual | Medium to high | Mixed evidence | Covert exploitation of vulnerability and cognitive autonomy | Restrictions on sensitive inference, reward audits, and meaningful exit |
| 9. Conversational entrapment and recruitment | Operator | conversation | Individual | Medium | Documented harms and demonstrated scaling | Dependency, exploitation, recruitment, or crisis amplification | Disclosure, age safeguards, crisis escalation, and trusted human support |
| 10. Deepfake psychological operations | Tool | synthetic media | Individual, institution, or population | Low | Documented current use | False evidence, impersonation, crisis action, and generalized distrust | Trusted channels, out-of-band verification, and crisis response |
| 11. Predictive population management | Environment | prediction | Population, group, place, or individual | Medium environmental authority | Documented current use; accuracy context dependent | Biased prediction, surveillance, and preemptive intervention | Aggregate design, uncertainty, privacy, due process, and audit |
| 12. AI psychological authority | Environment | authority | Individual or institution | High perceived authority | Demonstrated behavior; prevalence incomplete | Epistemic dependence, emotional displacement, and hidden institutional power | Uncertainty, source transparency, bounded roles, and human escalation |
Relationship map
Categories overlap, but overlap is not proof of combined use.
Every connection is labeled as documented, conceptual, environmental, organizational, or prospective. The map avoids implying that every technically possible combination has occurred.
Documented overlap: Generated text and media can supply high-volume assets to coordinated distribution networks.
Conceptual dependency: Persistent synthetic identities can provide the social interface through which autonomous agents operate.
Common overlap: Personalization can become manipulation when a system covertly exploits inferred vulnerability or adapts pressure in real time.
Documented overlap: Synthetic audio, video, and images can serve as propaganda assets or fabricated evidence.
Potential combination: A swarm can amplify a synthetic artifact, but the taxonomy does not assume every swarm contains deepfakes.
Environmental dependency: Ranking and recommendation systems determine whether coordinated content remains marginal or reaches wider audiences.
Prospective relationship: Population forecasts could inform timing or placement of interventions, but broad operational use is not presumed.
Documented overlap: A conversational system can become an emotional or epistemic authority as dependency and trust deepen.
Organizational relationship: Generated propaganda may be one product-development component inside a human-led traditional campaign.
Editorial-verification update
All sources resolved; every substantive claim linked.
The twelve supplied reports remain the primary synthesis sources. All 91 selected bibliography records have resolution entries and are referenced by the 501-claim matrix and 4,157-edge citation graph. Thirty high-impact sources received bounded automated retrieval or authoritative-source substitution. The default v100.0.46 package has no authorized review key and no imported signed human-review artifact; human editorial review, publication acceptance, and full-corpus independent re-verification remain false.
Source resolution
91 of 91 records
Every selected reference has a publisher, URL, resolution state, category links, and claim links. Unresolved metadata remains explicit.
Open the complete registryIndependent source checks
30 of 91 references
High-impact records were retrieved, officially corroborated, or replaced by a stronger source. This is not full-corpus verification.
Inspect checked sourcesClaim traceability
501 linked claims
Every substantive category statement maps to selected report references; orphan claims are rejected by validation.
Browse the matrixCitation graph
4,157 traceability edges
Category-to-claim and claim-to-source links are generated from one canonical graph. Every claim and selected source participates.
Explore the graphEvidence tensions
10 visible tensions
Conflicting findings, source-authority issues, detection bias, and legal currentness remain explicit.
Read the tension registerEditorial acceptance
Human review pending
No named reviewer, decision, timestamp, or signature is recorded. The packet cannot be treated as publication acceptance.
Open the pending packetMethodology and evidence standards
How to read this research
This taxonomy synthesizes twelve supplied interdisciplinary reports. It preserves their distinctions among historical precedent, confirmed real-world use, demonstrated capability, emerging risk, contested evidence, and speculation. The release did not independently reverify time-sensitive external claims.
Interpretive rules
- Engagement, impressions, reach, virality, or account count are not treated as proof of persuasion or behavior change.
- Attribution is presented as confirmed, alleged, inferred, contested, or unknown according to the reports.
- Fluent output does not establish strategic competence, consciousness, or autonomy.
- Detection indicators are suggestive unless the underlying evidence supports a stronger conclusion.
- Case studies demonstrate possibility or documented use; they do not automatically establish prevalence.
- Legal interpretations are summarized from the reports and may vary by jurisdiction or change over time.
- Operational details are abstracted into capability, risk, detection, and governance concepts.
Evidence legend
- Documented current use
- The report identifies public evidence that the capability was used in a real setting.
- Demonstrated capability
- The capability was shown in a controlled study, prototype, or bounded deployment.
- Emerging capability
- Components exist, but reliable or broad real-world use remains limited.
- Primarily prospective
- The report treats the capability mainly as a future or conceptual risk.
- Evidence contested or incomplete
- Evidence, effect size, attribution, or interpretation is disputed or methodologically limited.
- Mixed evidence
- Different parts of the category have different maturity or evidentiary support.
Evidence and publication boundary
The twelve supplied reports remain the primary synthesis sources. All 91 selected bibliography records have resolution entries and are referenced by the 501-claim matrix and 4,157-edge citation graph. Thirty high-impact sources received bounded automated retrieval or authoritative-source substitution. The default v100.0.46 package has no authorized review key and no imported signed human-review artifact; human editorial review, publication acceptance, and full-corpus independent re-verification remain false.
The 12 complete reports are retained in a server-denied provenance directory. Public pages expose non-operational summaries and selected source links.
From capability awareness to societal resilience
No single detector or disclosure can solve an information-system problem.
The reports converge on layered, rights-respecting defenses. These principles reduce risk; none is a guarantee.
Media and information literacy
Teach people to verify source, timing, context, and evidence without treating every unfamiliar account or synthetic artifact as malicious.
Source verification and provenance
Use trusted channels, signatures, content credentials, and out-of-band confirmation while recognizing that provenance can be absent, stripped, or spoofed.
Platform transparency
Explain ranking, enforcement, and coordinated-operation decisions and preserve privacy-protective researcher access.
Personal-data protection
Minimize collection, restrict sensitive inference, and prevent targeting based on distress, age, disability, grief, or financial crisis.
Human review for high-impact decisions
Do not allow opaque model scores or synthetic authority to determine employment, policing, health, education, or legal outcomes without accountable human review.
Crisis verification procedures
Pre-establish trusted channels, evidence preservation, rapid triage, and truth-first public communication.
Protection for vulnerable users
Use age-appropriate defaults, clear AI disclosure, crisis escalation, and safeguards against coercive dependency or sexual and financial exploitation.
Accountability for automated agents
Define responsibility across providers, developers, deployers, platforms, and operators; bound permissions and retain proportionate audit evidence.
Careful detection and attribution
Combine multiple signals, communicate confidence, provide appeals, and avoid false accusations based on a detector score alone.
Institutional trust and correction
Correct quickly, distinguish confirmed facts from inference, disclose uncertainty, and avoid sensationalizing adversary capability.
Continue with category-level evidence.
Each detailed page separates definition, confirmed use, demonstrated capability, plausible development, speculation, measured effects, unknowns, failure modes, defensive indicators, safeguards, research gaps, and selected sources.