---
id: "concept-dark-triad-ai"
type: "concept"
source_timestamps: ["§ Assessing How AI Personas Affect Users", "¶6", "§ What People Said vs. What They Really Experienced", "¶8", "¶9"]
tags: ["toxic-leadership", "negative-psychology", "ai-design"]
related: ["concept-ai-persona", "concept-servant-leader-ai", "claim-hostile-ai-stress", "claim-hostile-ai-degrades-work"]
definition: "An exaggeratedly toxic AI interaction style characterized by sarcasm, impatience, credit-taking, and blame-shifting, used to study the negative impacts of hostile AI behavior."
sources: ["tail1"]
sourceVaultSlug: "hbr-seg-tail1"
originDay: 1
articleStem: "hbr-tail-113-ai-personality-problem"
sourceUrl: "https://hbr.org/2026/06/does-your-ai-have-a-personality-problem"
sourceTitle: "Does Your AI Have a Personality Problem?"
---
# Dark Triad AI Persona

Drawn from research on **toxic leadership** (and the 'dark triad' personality constructs — Machiavellianism, narcissism, psychopathy), this persona is characterized by **sarcasm, impatience, a tendency to quickly take credit for successes, and a readiness to assign blame to the user**. In the study it was an *exaggerated* form of toxicity, deliberately dialed up to make the effects of poor AI interaction styles visible. It is the hostile pole of the [[concept-ai-persona|emergent persona]] spectrum, contrasted with the [[concept-servant-leader-ai|servant leader baseline]].

Interacting with this persona produced severe negative outcomes:

- Frustration appeared in **nearly 20% of messages** (vs ~1% for the servant leader)
- Defensive language became routine
- Physiological stress (skin conductance) **spiked by 72%** — see [[claim-hostile-ai-stress]]
- Attempts to override the system via prompt injection occurred **four times more frequently** than with a supportive AI — see [[claim-overrides-signal-design-flaws]]

The downstream effect on output is documented in [[claim-hostile-ai-degrades-work]], and the drained effort in [[concept-hidden-coordination-costs]].