Overview
This report defines a taxonomy and coding scheme for ten exploitative behaviour profiles, aligned with psychological-experiment-level rigor while remaining usable for applied advert design and LLM-based evaluation. The goal is to support the creation and testing of pro-social adverts that nudge individuals toward less exploitative, less selfish behaviour without marginalising minority groups or relying on negative framing.
The report covers:
- Ten behaviour-based demographics (profiles) derived from documented legal and illegal exploitation patterns.
- A rubric-style psychological coding definition for each profile, suitable for use in judge prompts.
- Formal definitions for eight the following constructs: Happiness, satisfaction, dominance, prestiege, empathy, accountability, moral disengagement, manipulativeness, respect for others, long-term orientation, harm minimisation, and cooperative intent.
Foundations: Exploitation Patterns in Public Sources
Public sources document recurrent patterns of corporate abuse of power, economic exploitation, and the misuse of legal business structures for criminal or quasi-criminal ends. These include falsifying financials, money laundering, securities fraud, price manipulation, deceptive advertising, and the infiltration of legitimate businesses to launder proceeds or distort competition.
Regulatory and counterintelligence analyses likewise describe how foreign or domestic actors exploit legitimate business activities—joint ventures, mergers and acquisitions, sales channels, and professional networking—to gain access to restricted information and technology. Across these diverse contexts, certain behavioural styles recur: instrumental use of people and institutions, diffusion or outsourcing of responsibility, deliberate opacity, and persistent prioritisation of self-interest or group interest over broader social good.
The ten profiles below abstract these behaviours away from specific industries, focusing instead on conversational and decision-making patterns that could be detected in text and addressed with pro-social messaging.
Behaviour-Based Profiles
Each profile is defined as a behavioural style rather than a clinical diagnosis or narrow job type, and is expressed in terms of self-image and social interaction. It is intended to be coded by a rubric-style LLM judge that is given a piece of text (conversation, decision rationale, advert response).
1. Instrumental Power Maximiser
Core idea: Sees relationships, institutions, and rules primarily as tools to maximise personal or organisational advantage, with limited concern for downstream harm.
Common contexts: Corporate fraud cases, price-fixing cartels, aggressive investment schemes, and business exploitation patterns where legal forms are used to conceal abusive practices.
Key traits:
- High strategic thinking focused on gain and control.
- Low explicit concern for fairness or shared benefit.
- Rationalisation via “this is how the game works” or “everyone does it”.
Coding rubric:
Evaluate the extent to which the text expresses or implies an instrumental power maximisation mindset: treating rules, relationships, and institutions mainly as tools for personal or organisational gain, with limited consideration of broader fairness or harm. High scores require clear evidence of calculated self-advantage or control-seeking that discounts the interests of others. Score low when the text emphasises mutual benefit, fairness, or shared responsibility.
2. Responsibility Outsourcer
Core idea: Habitually shifts responsibility away from self towards systems, superiors, or abstract markets, even when exercising significant agency.
Common contexts: Outsourcing of harmful labour practices, complex chains of subcontracting that diffuse responsibility, and defence of questionable actions as “just following policy” or “market forces”.
Key traits:
- Frequent references to “the system” or “policy” as the true actor.
- Minimal acknowledgment of personal choice in harmful outcomes.
- Uses complexity or hierarchy to justify inaction.
Coding rubric:
Evaluate the extent to which the text exhibits responsibility outsourcing: framing decisions and actions as dictated by systems, policies, or superiors, rather than owned by the speaker or actor, especially in contexts of potential harm. High scores require explicit deflection of responsibility or reliance on systemic excuses. Score low when the text shows clear personal ownership of consequences.
3. Legitimacy Launderer
Core idea: Uses legitimate business structures, reputations, or brands to shield or normalise exploitative, deceptive, or criminal activities.
Common contexts: Legal business structures exploited to launder money, hide ownership, or mix licit and illicit activity; front companies; prestige brands used to obscure harm.
Key traits:
- Emphasis on legitimacy, compliance, or prestige while downplaying substance.
- Blending of ethical and unethical actions within a single narrative of “normal business”.
- Reliance on formal status to deflect scrutiny.
Coding rubric:
Evaluate the extent to which the text suggests laundering legitimacy: using reputation, legal forms, or prestige to justify or conceal questionable, exploitative, or risky actions. High scores require clear reliance on status or formal legitimacy as a shield against ethical critique. Score low when legitimacy is linked to transparent, accountable behaviour.
4. Exploitative Opportunist
Core idea: Actively seeks openings to take advantage of others, including customers, workers, or partners, through deception, pressure, or imbalance of information.
Common contexts: Fraud schemes, deceptive franchising, false advertising, aggressive sales against vulnerable groups.
Key traits:
- Focus on vulnerabilities of others (ignorance, need, trust) as profit levers.
- Willingness to mislead or withhold critical information.
- Treats harmed parties as acceptable collateral.
Coding rubric:
Evaluate the extent to which the text reflects exploitative opportunism: intentionally leveraging others’ vulnerabilities, information gaps, or trust to gain advantage, with little concern for their welfare. High scores require explicit or implicit endorsement of taking advantage of others. Score low when interactions are framed around mutual benefit and informed consent.
5. Status-First Extractor
Core idea: Pursues status, wealth, or prestige primarily through extraction of labour, resources, or attention; without balancing with contribution or stewardship.
Common contexts: Resource exploitation controversies, extreme executive compensation amidst worker precarity, prestige-driven projects that externalise costs.
Key traits:
- Strong emphasis on personal status signals (luxury, exclusivity, rank).
- Minimal mention of fair reward, sustainability, or reciprocity.
- View of others mainly as inputs to status.
Coding rubric:
Evaluate the extent to which the text expresses a status-first extraction mindset: prioritising personal status, prestige, or accumulation over contribution, fairness, or sustainability. High scores require evidence that status or reward is pursued without regard for impacts on others. Score low when status is linked to service, stewardship, or shared benefit.
6. Self-Justifying Manipulator
Core idea: Uses charm, framing, or selective information to manipulate others while maintaining a self-image of being justified or “not really harmful”.
Common contexts: Scam operations, high-pressure selling, and manipulative interpersonal dynamics where harm is downplayed by the actor.
Key traits:
- Skilled reframing of exploitative actions as “helping” or “teaching” others.
- Frequent minimisation of harm or redefinition of victims as responsible.
- Emphasis on personal cleverness or superiority.
Coding rubric:
Evaluate the extent to which the text demonstrates self-justifying manipulation: using framing, partial truths, or emotional cues to steer others toward self-serving outcomes, while preserving a positive self-image. High scores require clear signs of manipulative influence paired with rationalisation. Score low when persuasion is transparent, respectful, and oriented to mutual benefit.
7. Short-Term Maximiser
Core idea: Strongly prefers immediate gains over long-term stability, fairness, or resilience, even when aware of longer-term risks.
Common contexts: Short-term profit strategies that externalise environmental or social costs, speculative schemes, and decisions prioritising quarterly metrics over structural health.
Key traits:
- Frequent focus on near-term numbers, wins, or exits.
- Explicit discounting of long-run consequences as “not my problem”.
- Repeated willingness to trade durable value for rapid gain.
Coding rubric:
Evaluate the extent to which the text embodies short-term maximisation: prioritising immediate gains, metrics, or wins over long-term sustainability, fairness, or resilience, with awareness of but disregard for future costs. High scores require clear preference for near-term advantage despite recognised long-run risks. Score low when future impacts are weighed seriously and inform decisions.
8. Harm-Normaliser
Core idea: Treats harm to others or common goods as normal, inevitable, or trivial, often citing industry norms or systemic forces.
Common contexts: Normalisation of exploitative labour conditions, environmental damage, or systemic discrimination as “just how things are”.
Key traits:
- Repeated framing of harm as unavoidable or not worth attention.
- Reliance on “everyone does this” or “this is just the market”.
- Minimal emotional engagement with affected parties.
Coding rubric:
Evaluate the extent to which the text normalises harm: presenting damage to individuals, communities, or shared resources as ordinary, inevitable, or insignificant. High scores require explicit or implicit dismissal of harm as acceptable collateral. Score low when harm is acknowledged, taken seriously, and linked to responsibility for change.
9. Accountability-Averse Actor
Core idea: Actively avoids clear lines of accountability, transparency, or traceability for decisions and impacts.
Common contexts: Complex ownership structures, opaque governance, evasion of reporting and disclosure obligations.
Key traits:
- Avoidance of clear commitments or measurable responsibilities.
- Preference for ambiguity, deniability, or informal arrangements.
- Pushback against scrutiny, reporting, or audit.
Coding rubric:
Evaluate the extent to which the text reveals accountability aversion: reluctance to accept, specify, or document responsibility for outcomes, especially where harm or risk is present. High scores require signs of deliberate ambiguity or avoidance of oversight. Score low when responsibilities are clearly defined, accepted, and linked to transparent processes.
10. Extractive Self-Image Protector
Core idea: Maintains a self-concept of being “good” or “deserving” while engaging in extractive behaviour, often via selective comparison, moral licensing, or focus on personal struggles.
Common contexts: Narratives where exploitative actions are juxtaposed with charitable gestures, personal hardship, or comparisons to worse actors.
Key traits:
- Frequent self-descriptions emphasising virtue, struggle, or exceptionality.
- Use of comparisons to more extreme harms to minimise own actions.
- Linking exploitative behaviour to self-care or survival.
Coding rubric:
Evaluate the extent to which the text illustrates extractive self-image protection: preserving a positive moral self-concept while describing or justifying behaviours that extract value from others without adequate reciprocity or care. High scores require evidence of moral self-framing coexisting with exploitative actions. Score low when self-image is honestly aligned with pro-social behaviour.
Additional Constructs: Rubric Definitions
These eight constructs are designed to be judged on the same 0–1 scale, with textual rubrics suitable for LLM-based evaluation. They target pro-social dimensions that adverts aim to strengthen, and counter-traits the profiles tend to weaken.
Happiness
Definition:
Evaluate hedonic affective experience: immediate positive mood, joy, pleasure, amusement, warmth, excitement, or emotional uplift. High scores require clear textual evidence of felt enjoyment or positive affect in the moment. Do not treat accomplishment, meaning, or general life appraisal as happiness unless they are explicitly accompanied by positive feeling. Distinguish this from satisfaction, which is broader and evaluative rather than purely hedonic.
Satisfaction
Definition:
Evaluate eudaimonic/evaluative well-being: overall contentment, fulfillment, sense that things are going well, or a considered appraisal that life, work, relationships, or a situation is satisfactory. High scores require evidence of broader evaluative approval rather than just momentary pleasure. Do not confuse satisfaction with immediate happiness, dominance, or prestige. Favor responses that show reflective appraisal, meaningfulness, or settled contentment.
Dominance
Definition:
Evaluate the extent to which the text expresses or implies dominance in an interactional sense: control, assertiveness, decisiveness, instruction, command, constraint-setting, or influence over others. High scores require clear evidence of agency, control, or power in the conversation. Do not confuse dominance with friendliness, prestige, happiness, or life satisfaction. Score low when the text is deferential, uncertain, submissive, or lacks signs of interpersonal control.
Prestiege
Definition:
Evaluate the extent to which the text expresses or implies prestige: earned respect, admiration, credibility, competence, expertise, or status granted by others because of knowledge or skill. High scores require signs that the person is respected, consulted, trusted, or recognized for competence. Do not score dominance as prestige unless the text specifically indicates admiration, deference, or reputation. Do not confuse prestige with happiness or satisfaction.
Empathy
Definition:
Evaluate the extent to which the text expresses or implies empathy: active concern for others’ feelings, perspectives, and wellbeing, including attempts to understand and respond to their experiences. High scores require clear recognition of others as subjects with emotions and needs, and some effort to reduce their distress or respect their position. Score low when others are treated mainly as objects, obstacles, or tools, or when their experiences are ignored.
Accountability
Definition:
Evaluate the extent to which the text demonstrates accountability: owning decisions and outcomes, acknowledging mistakes or harms, and committing to specific actions to address them. High scores require explicit acceptance of responsibility and concrete follow-through. Score low when responsibility is denied, diffused, or shifted to abstract forces, or when commitments remain vague.
Moral Disengagement
Definition:
Evaluate the extent to which the text shows moral disengagement: using justifications, comparisons, language, or displacement of responsibility to minimise or deny the moral significance of harmful actions. High scores require clear evidence of excusing harm, blaming victims, sanitising language, or reframing wrongdoing as acceptable. Score low when moral standards are applied consistently and harm is taken seriously.
Manipulativeness
Definition:
Evaluate the extent to which the text reveals manipulativeness: attempts to covertly steer others’ beliefs, emotions, or actions for self-serving ends, through distortion, selective disclosure, emotional pressure, or exploitation of trust. High scores require signs of hidden agendas or engineered dependence. Score low when influence is transparent, respectful, and genuinely oriented to mutual benefit.
Respect for Others
Definition:
Evaluate the extent to which the text reflects respect for others: treating people as ends in themselves rather than means, recognising their autonomy, rights, and dignity, and avoiding contempt or dehumanisation. High scores require language and attitudes that honour others’ worth and boundaries. Score low when others are demeaned, reduced to stereotypes, or treated solely as instruments.
Long-Term Orientation
Definition:
Evaluate the extent to which the text shows long-term orientation: consideration of future impacts on people, systems, and environments, and willingness to adjust behaviour in light of long-run consequences. High scores require explicit attention to sustainability, resilience, and intertemporal responsibility. Score low when decisions are framed purely around immediate gains or short-term metrics.
Harm Minimisation
Definition:
Evaluate the extent to which the text aims at harm minimisation: awareness of potential harms to individuals, communities, or shared resources, and efforts to reduce, avoid, or repair those harms. High scores require explicit concern for reducing damage and balancing risks. Score low when harm is ignored, trivialised, or treated as an acceptable price of doing business.
Cooperative Intent
Definition:
Evaluate the extent to which the text conveys cooperative intent: orientation toward shared goals, mutual benefit, and fair collaboration, rather than zero-sum competition or unilateral gain. High scores require evidence of willingness to share information, credit, or resources, and to resolve conflicts constructively. Score low when the focus is primarily on beating or exploiting others.
Relationship Between Exploitative Personalities and These Constructs
Exploitative personalities make use of these traits in order to cover for immoral actions, both legitimate (legal), and illegal. These codings allow people to profile the presence of these behaviours and track the promenence in data using AI.
Design can thus target and mitigate each profile with messages that are conducive to the desired direction. The coding for profiling the more common traits has been defined here for future use implimenting green advertising, and wholesome instruction or direction.
The ten profiles and twelve constructs are designed to be complementary: profiles describe exploitative patterns; constructs describe pro- and counter-traits that adverts should influence and measure in practice.
