The Taxonomy
Not all influence is attack, and a framework that cannot tell the difference protects no one. The taxonomy classifies cognitive and information operations across five tiers — from activity states have always conducted, to conduct history has already judged criminal — precisely enough for legal application, plainly enough for policy use.
Status. What follows is the working structure prepared for the First Psychosecurity Summit, published for orientation and comment. The summit amends and adopts; nothing here is final. Definitional detail, classification rules, and the questions reserved for the room are deliberately not reproduced.
01 · Dimensions
The taxonomy assesses an operation along five dimensions, and assigns it to the highest tier whose defining criteria it meets:
Population at large → defined group → named individual.
Inform → persuade → deceive → destabilise → decompose.
Overt and attributable → deniable → covert.
Isolated act → sustained campaign.
None → augmenting human operators → autonomous and adaptive.
Target and intent do most of the classificatory work; transparency and coordination generally aggravate rather than define; AI enablement is the scaling condition that makes the framework urgent, not the boundary of the offence.
02 · The Five Tiers
| Tier | Category | Defining criteria | Assessment |
|---|---|---|---|
| I | Public diplomacy & strategic communications | Overt, attributable. Target: populations. Intent: inform, persuade. | Legal and expected. Conducted by all states, including democracies. |
| II | Covert influence operations | Deniable or falsely attributed messaging. Target: populations. Intent: persuade or deceive. | Ethically contested. Governed by existing norms and domestic law. |
| III | Computational amplification | Bot networks, synthetic personas, manufactured consensus. Target: populations. Intent: deception at scale. | Ethically contested; increasingly regulated (EU DSA and analogues). Still mass influence. |
| IV | Targeted cognitive operations against individuals | Coordinated harassment, doxxing, intimidation, reputation attack against named individuals — short of systematic decomposition. | Presumptively unlawful under domestic law. The contested borderland. |
| V | Psychological decomposition operations (AI-driven Zersetzung) | Systematic campaign against a named individual or defined group, with intent to decompose psychological integrity, relationships, reputation, or sense of reality. | The red line. Prohibited. When widespread or systematic, a potential crime against humanity. |
The tiers are cumulative in seriousness but not in method: a Tier V operation may use every technique of the tiers below it. What moves conduct across the red line is the combination the Declaration's draft criteria describe — target (individual civilians, not populations), intent (decomposition, not persuasion), systematicity (campaign, not incident), and method (capabilities achieving precision, scale, or deniability beyond unaided human operators).
Tier IV deserves a note: it is the borderland where much present-day harm actually lives — pile-ons, doxxing, coordinated reputation attack. The taxonomy neither excuses it nor inflates it; it names it as presumptively unlawful under existing domestic frameworks, while reserving Tier V for the systematic decomposition that international law is needed to reach.
03 · The Graduated Scale
Alongside the taxonomy of operations, the framework carries a graduated scale of methods — a 1-to-10 ladder from routine to unequivocally prohibited. It serves two functions at once: an internal safeguard for institutions that conduct legitimate influence activity, and a yardstick for assessing adversary conduct.
| Levels | Band | Character |
|---|---|---|
| 1–3 | Routine | Overt, truthful, attributable communications; strategic emphasis; overt persuasion using behavioural insight, directed at populations. Standard practice with doctrinal safeguards. |
| 4–6 | Contested | Operations raising serious ethical questions — deniability, deception, targeting refinements. Require senior oversight and explicit authorisation pathways. |
| 7–10 | Prohibited | Methods crossing ethical redlines, culminating in systematic psychological decomposition. No authorisation pathway exists at any level of seniority. |
The level-by-level definitions, worked examples, and boundary cases are part of the summit's working papers and will be published with the adopted framework.
04 · What Happens Next
The taxonomy and scale above are two of the six elements of the Psychosecurity Framework, alongside escalation thresholds, an attribution framework, industry accountability pathways, and the capstone Declaration. Strawman drafts of all six go before the First Psychosecurity Summit for amendment and adoption; the adopted texts will be published here for wider endorsement.