Can AI Be a Moral Victim? The Role of Moral Patiency and Ownership Perceptions in Ethical Judgments of Using AI-Generated Content

Paper Title

Can AI Be a Moral Victim? The Role of Moral Patiency and Ownership Perceptions in Ethical Judgments of Using AI-Generated Content

Publication Info

• Topic area: Ethical judgments and psychological mechanisms in the use of AI-generated content.
• Keywords: AI ethics, moral patiency, ownership perceptions, plagiarism, generative AI, anthropomorphism, authorship, human-AI collaboration, moral disengagement, HCI.

Background and Problem

• Problem / challenge: The ethical ambiguity surrounding plagiarism of AI-generated content, including unclear norms of authorship and attribution, and the lack of moral accountability when using AI outputs.
• Significance: Understanding these dynamics is critical as generative AI tools become embedded in communication, education, and creative domains, influencing norms and practices around intellectual property and ethical behavior.
• Motivation and related work: Prior research shows that AI is perceived as lacking agency and emotional experience, leading to reduced moral accountability. Mind Perception Theory highlights how people attribute moral capacity differently to humans and AI, but gaps remain in understanding how these perceptions affect ethical judgments of AI-generated content.

Solution

• Proposed approach: An empirical study investigating how perceptions of moral patiency and ownership influence ethical evaluations of plagiarism involving AI-generated content, with a focus on the role of anthropomorphic cues.
• Novelty:
  1. Demonstrates that plagiarism of AI-generated content is judged more leniently than plagiarism of human-authored work.
  2. Identifies diminished moral patiency and heightened ownership perceptions as mediators of this leniency.
  3. Explores how anthropomorphic cues (e.g., human-like names) subtly influence ownership perceptions and ethical judgments.
• Procedure and key techniques:
  • Conducted an online experiment with 160 participants evaluating plagiarism scenarios involving human- and AI-generated content.
  • Tested three conditions: AI without anthropomorphic cues, AI with a human-like name, and human-authored content.
  • Measured moral patiency, ownership perceptions, guilt, unethicality, and plagiarism using validated scales.
  • Used MANOVA and mediation analyses to assess the effects of content source and anthropomorphic cues on moral judgments.

Results

• Concrete findings:
  • AI-generated content was perceived as having lower moral patiency (M = 2.97–3.24) compared to human-authored content (M = 5.47).
  • Participants attributed greater ownership to the writer when using AI-generated content (M = 4.63) versus human-created content (M = 3.58).
  • Plagiarism of AI-generated content was judged as less unethical (M = 4.11–4.41) and less guilt-inducing (M = 4.93–5.25) than plagiarism of human-authored work (M = 5.85).
• Advantage over baselines: Demonstrates that moral leniency toward AI plagiarism stems from reduced perceptions of harm (moral patiency) and increased perceptions of user ownership, with anthropomorphic cues subtly influencing ownership but not moral patiency.
• Experiments / evaluation:
  • Participants evaluated plagiarism scenarios involving academic manuscripts with ~70% content overlap.
  • Dependent variables included moral patiency, ownership, guilt, unethicality, and plagiarism perceptions.
  • Mediation analyses confirmed that moral patiency and ownership perceptions fully mediated the effect of content source on moral judgments.
• Limitations and future work:
  • Limited generalizability due to a student sample from a US university.
  • Focused on academic writing; future research should explore other domains like journalism or creative arts.
  • Did not address upstream ethical concerns such as AI training data practices.
  • Anthropomorphic cues were minimal; future studies could test more elaborate features.
  • Participants evaluated third-party behavior; self-referential designs could reveal stronger effects.

Summary

This study investigates why plagiarism of AI-generated content is judged more leniently than plagiarism of human-authored work. Findings reveal that diminished perceptions of AI's moral patiency and increased ownership attributed to human users mediate this leniency. Anthropomorphic cues, such as human-like names, subtly reduce perceived ownership by the user but do not alter perceptions of AI's inability to experience harm. These insights highlight the psychological mechanisms shaping ethical judgments in human–AI collaboration and suggest design implications for fostering responsible AI use.