Educational companion dossier · Fact, interpretation, lived experience, clinical education, fiction, and mechanics are labeled separately. Scope & safety
PsychologicalWar.orgHistory before hype.

AI PSYOPS TAXONOMY · CATEGORY 10

AI-Enabled Deepfake Psychological Operations

Synthetic or manipulated audio, video, imagery, or multimodal evidence is used to impersonate, provoke, discredit, or undermine trust for an influence objective.

Primary level: Tool Documented current use Claim AIP-10-A01
Read evidence assessment All 12 categories

CURRENT EVIDENCE ASSESSMENT

Synthetic audio, image, and video use is documented; deception, confusion, and liar’s-dividend effects depend more on context than perfect realism alone.

Stable claim identifierAIP-10-A01
Claim stageassessment
Currentness reviewed2026-07-27T00:00:00Z

DEPLOYMENT

Documented

Election, fraud, harassment, and influence incidents involving synthetic media are publicly established.

AUTONOMY

Tool-level production

Humans typically select the target, narrative, timing, and distribution channel.

PERSISTENCE

Episodic asset; persistent epistemic effects possible

A specific asset may be brief while uncertainty and denial can outlast it.

PROFILING ACCURACY

Not required

Success can rely on identity, timing, confirmation bias, and information vacuums rather than individual profiles.

MEASURED EFFECT

Confusion and judgment effects demonstrated; broad behavior variable

Experiments and incidents show deception or doubt, but not uniform persuasion or action.

Assessment basis

Assessment combines the exact owner-supplied category report with the bounded primary, official, platform, and peer-reviewed sources listed for this category. Dimensions are evaluated separately to prevent documented output from being mistaken for autonomy or effect.

What would change this assessment

Upgrade claims with verified exposure, authentication timelines, response delays, and measured downstream decisions.

Prohibited inference

Do not infer strategic effect, universal deployment, or individual psychological state from this assessment.

A · DEFINITION

What this category means sources

Definition

A deepfake psychological operation intentionally uses synthetic or manipulated audio, video, imagery, or multimodal media to alter perception, impersonate a trusted figure, create false evidence, provoke action, discredit authentic material, or undermine trust.

Outside this category

Not all edited media is a deepfake, and not all deepfakes are psychological operations. Disclosed artistic transformation, accessibility dubbing, satire, and ordinary editing lack the deceptive influence objective that defines this category.

B · SIGNIFICANCE

Why it matters sources

A synthetic artifact released during a crisis can outrun verification. The broader danger is epistemic: once people know convincing fakes exist, wrongdoers can deny authentic recordings and institutions may struggle to establish a shared factual record.

C · CHANGE FROM PRE-AI PRACTICE

How AI changes the phenomenon sources

Generative models make voice, face, image, and event fabrication cheaper and faster. Success often depends more on timing, source credibility, prior belief, and emotional intensity than on perfect realism. Low-quality manipulations can still work in an information vacuum.

D · CAPABILITY STATUS

Separate evidence from projection sources

DOCUMENTED

Confirmed real-world use

Synthetic audio, video, and imagery have been used in political, criminal, and coercive incidents.

DEMONSTRATED

Demonstrated technical capability

Current tools can impersonate voices and faces and generate plausible scenes, though artifacts and temporal errors remain.

EMERGING

Plausible near-term development

Real-time multimodal impersonation and hybrid authentic-synthetic evidence are likely to become more accessible.

UNCERTAIN

Unsupported or unproven

Technical realism does not establish persuasion, and no detector can provide universal certainty across all media and generation methods.

E · KEY MECHANISMS

Conceptual mechanisms — not an operating procedure sources

Safety transformation: these descriptions identify system functions at a high level. Procedural steps, target criteria, scripts, evasion methods, and deployment workflows are intentionally excluded.

  1. Voice cloning and fabricated calls or recordings.
  2. Face replacement, reenactment, and altered lip synchronization.
  3. Synthetic eyewitness, battlefield, protest, or disaster imagery.
  4. Hybrid fakes that change one consequential element inside authentic media.
  5. False claims that genuine evidence is AI-generated.

F · EVIDENCE & EXAMPLES

What is known, measured, and still unknown sources

REACH IS NOT EFFECT. Publication, impressions, engagement, virality, or media attention do not by themselves establish persuasion or behavioral change.

Example 1 · Claim AIP-10-E01

Election-related synthetic audio

Synthetic media documented; behavioral impact uncertain

What occurred
Fabricated political audio circulated during election periods, including cases described in Slovakia and the United States.
What is confirmed
The media artifacts and distribution are documented.
Effect measured
They triggered public warnings, investigation, and policy response.
What remains unknown
A reliable causal effect on final voter behavior is not established.
Source scope
The linked sources support the bounded statements shown here; they do not automatically establish intent, reach, persuasion, behavior, or strategic effect.
Correction trigger
Revise when a primary record, authoritative correction, adjudication, retraction, or stronger causal study changes the bounded statement.
Linked sources: SRC-17-FCC-NH-DEEPFAKE-ROBOCALL Federal Communications Commission
Example 2 · Claim AIP-10-E02

Wartime leader-impersonation media

Public incident reporting exists; strategic effect not established

What occurred
Synthetic or manipulated media impersonating political leaders circulated during armed conflict and information crises.
What is confirmed
Specific artifacts and distribution events have been documented by investigators and journalists.
Effect measured
Rapid debunking and limited visible engagement were reported in some cases; no broad strategic effect is established.
What remains unknown
Complete exposure, belief, delayed influence, and the independent effect of the synthetic asset remain unknown.
Source scope
The linked sources support the bounded statements shown here; they do not automatically establish intent, reach, persuasion, behavior, or strategic effect.
Correction trigger
Revise when a primary record, authoritative correction, adjudication, retraction, or stronger causal study changes the bounded statement.
Linked sources: SRC-23-CHESNEY-CITRON-DEEPFAKES California Law Review SRC-07-GRAPHIKA-WOLF-NEWS Graphika
Example 3 · Claim AIP-10-E03

The liar’s dividend

Legal theory and experimental evidence; real-world magnitude context-dependent

What occurred
Researchers and legal scholars examine how awareness of deepfakes can make false denials of authentic evidence more plausible.
What is confirmed
The mechanism is theoretically developed and has experimental support under tested conditions.
Effect measured
Some participants become more uncertain or receptive to false denial claims.
What remains unknown
Its prevalence, durability, and ability to override trusted authentication in real crises remain unsettled.
Source scope
The linked sources support the bounded statements shown here; they do not automatically establish intent, reach, persuasion, behavior, or strategic effect.
Correction trigger
Revise when a primary record, authoritative correction, adjudication, retraction, or stronger causal study changes the bounded statement.
Linked sources: SRC-25-SCHIFF-LIARS-DIVIDEND Psychology of Popular Media SRC-24-LUCAS-DEEPFAKE-EFFECTS Journal of Computer-Mediated Communication

G · RISKS & FAILURE MODES

Potential harms and reasons the capability may fail sources

Risks

  • Rapid false evidence can trigger panic, fraud, or escalation before verification.
  • Private individuals may lack the resources to rebut convincing impersonation.
  • False positives can discredit authentic media and innocent creators.
  • Provenance metadata can be removed, absent, or forged.
  • Repeated exposure to synthetic media can erode baseline trust even when specific fakes are debunked.

Limitations and failure modes

  • Long video can still reveal temporal, physical, acoustic, or contextual inconsistencies.
  • Human detection performs poorly for high-quality media and can be overconfident.
  • Detection models are generation-specific and degrade as tools change.
  • Labels and watermarks are incomplete evidence, not proof of authenticity or falsity.

H · DETECTION & DEFENSIVE INDICATORS

Signals for investigation, not automatic verdicts sources

Indicator rule: unless the source report supports a stronger conclusion, each signal below is suggestive rather than conclusive. Multiple independent signals and contextual evidence are required.

  • A sensational recording appearing first through an unverified channel should trigger source confirmation before rebroadcast.
  • Inconsistent lighting, anatomy, timing, acoustics, or context may be suggestive but not conclusive.
  • Absence of provenance should increase uncertainty, not automatically classify the media as fake.
  • Independent confirmation from the represented person or institution is often more valuable than automated scoring alone.

I · GOVERNANCE & SAFEGUARDS

Accountability, transparency, and human protection sources

Sign authoritative media and publish trusted verification routes before a crisis occurs.

Use layered forensic analysis and preserve original files and chain of custody.

Communicate uncertainty quickly and update public statements as evidence changes.

Protect private victims with rapid reporting, takedown, and legal-remedy processes.

Treat provenance, detectors, human review, and contextual verification as complementary—not interchangeable—controls.

J · RESEARCH GAPS

Questions the evidence does not yet close sources

  • Long-term effects on baseline trust and democratic participation.
  • Effectiveness and unintended consequences of mandatory labels.
  • Privacy-preserving provenance that remains useful after legitimate editing.
  • Crisis protocols for people and institutions with limited technical resources.
REAL-WORLD INTERPRETIVE

L · SOURCES & REVIEW STATUS

Exact owner report, claim register, and reviewed sources

  1. AI-Enabled Deepfake Psychological Operations Owner-supplied report: Deepfake Psychological Operations Research.md · 62,023 bytes · SHA-256 8dec5d4accf0bd6fea3d357e9648ae18d77b3939a7c628c1caf109ed2e47e09a

    Owner-supplied interdisciplinary research synthesis; exact source preserved in protected durable memory. External specialist review remains pending.

Claim-specific reviewed sources

  1. Federal Communications Commission · 2024-09-30 · Official public authority

    Supports
    Confirms an AI-generated voice-cloning robocall campaign targeting New Hampshire primary voters and the FCC enforcement response.
    Does not establish
    Does not establish how many recipients believed the message or whether it changed turnout; it does not independently verify separate election-audio incidents in other countries.
    Review
    LOCATED_AND_REVIEWED_AT_CITATION_LEVEL · Currentness checked for the bounded claim scope on 2026-07-27.

  2. California Law Review · 2019-12-01 · Peer-reviewed or editorially reviewed scholarship

    Supports
    Defines major deepfake risks, including impersonation, evidentiary disruption, and the liar’s dividend.
    Does not establish
    Legal analysis does not measure the prevalence or persuasive effect of specific incidents.
    Review
    LOCATED_AND_REVIEWED_AT_CITATION_LEVEL · Currentness checked for the bounded claim scope on 2026-07-27.

  3. Journal of Computer-Mediated Communication · 2022-06-30 · Primary research

    Supports
    Tests how political deepfake video affects deception, uncertainty, and trust under experimental conditions.
    Does not establish
    A controlled study does not establish uniform effects across crises, cultures, or high-stakes operational contexts.
    Review
    LOCATED_AND_REVIEWED_AT_CITATION_LEVEL · Currentness checked for the bounded claim scope on 2026-07-27.

  4. Psychology of Popular Media · 2026-01-01 · Primary research

    Supports
    Examines whether deepfake awareness can make false denials of authentic audiovisual evidence more credible.
    Does not establish
    Does not establish that every denial benefits from the liar’s dividend or that the effect dominates all verification cues.
    Review
    LOCATED_AND_REVIEWED_AT_CITATION_LEVEL · Currentness checked for the bounded claim scope on 2026-07-27.
  5. SRC-07-GRAPHIKA-WOLF-NEWSDeepfake It Till You Make It

    Graphika · 2023-02-07 · Independent specialist analysis

    Supports
    Documents limited use of AI-generated fictitious news presenters in content promoted by a pro-China influence operation.
    Does not establish
    Does not establish substantial reach or persuasive effect and should not be generalized to all synthetic presenter use.
    Review
    LOCATED_AND_REVIEWED_AT_CITATION_LEVEL · Currentness checked for the bounded claim scope on 2026-07-27.
Selected works identified by the owner-supplied report
  • Chesney and Citron, Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security.
  • Lucas et al., Political Deepfake Videos Are No More Deceptive Than Other Fake News.
  • Schiff et al., The Liar’s Dividend: Can Politicians Claim Misinformation to Evade Accountability?
  • Boneh, Using Zero-Knowledge Proofs to Fight Disinformation.

Exact source preservation and editorial currentness review do not constitute specialist certification, adjudication, legal advice, clinical review, or proof that every owner-report citation is current. Corrections remain open.

Evidence methodReach versus effectCorrectionsDefensive incident template

Page complete AI-Enabled Deepfake Psychological Operations Page label: CONTEMPORARY / ONGOING CLAIM — NOT SETTLED HISTORY