TLDR: A recent study involving 318 generative AI users found that ‘abuse anxiety’ significantly erodes trust, perceived usefulness, and the intention to use these technologies. Concerns about fraud, misinformation, and privacy violations, particularly from deepfake technology, are key drivers of this anxiety. The research highlights the critical need for enhanced transparency, accountability, and clear regulatory frameworks to foster user confidence and promote the sustainable development of generative AI.
New research published on August 14, 2025, in the Journal of Theoretical and Applied Electronic Commerce Research, sheds light on a critical barrier to the widespread adoption of generative artificial intelligence (AI): ‘abuse anxiety’ among users. The study, titled ‘When Generative AI Meets Abuse: What Are You Anxious About?’, surveyed 318 generative AI users and found a direct negative correlation between users’ concerns about potential misuse and their trust, perceived usefulness, and ultimate intention to use these advanced AI systems.
Generative AI, with its remarkable capabilities in creating realistic text, images, and audio, has rapidly transformed industries from entertainment to education. Tools like MidJourney, ChatGPT, and AIVA exemplify this technological leap, enabling tasks that once required specialized skills. However, this rapid advancement has also brought forth significant concerns regarding misuse, privacy risks, and ethical dilemmas.
The Rise of ‘Generative AI Abuse Anxiety’
The study introduces the concept of ‘Generative AI Abuse Anxiety’ (GAIA), defining it as users’ psychological concerns about generative AI being used for unethical, illegal, or harmful activities. This anxiety extends beyond traditional technology anxiety or general risk perception, specifically addressing fears related to online fraud, the spread of misinformation, and privacy violations. A prominent example cited is the malicious use of deepfake technology, which can forge identities, disseminate false information, or facilitate extortion, thereby eroding public trust.
Researchers Yuanzhao Song and Haowen Tan found that GAIA significantly undermines various dimensions of trust. Their findings indicate a negative impact on ‘Trust in Websites’ (concerning IT quality, security, and privacy), ‘Human-like Trust’ (pertaining to AI’s ability, benevolence, and integrity), and ‘System-like Trust’ (related to fairness, accountability, transparency, and explainability). While the direct negative impact of GAIA on perceived usefulness and acceptance was statistically observable, the study notes these effects were relatively small, suggesting that the primary pathway of anxiety’s influence is through the erosion of trust.
Trust as a Cornerstone for AI Adoption
The research adapted the widely recognized Technology Acceptance Model (TAM), replacing the ‘perceived ease of use’ component with a multidimensional trust framework. This adjustment acknowledges that while generative AI interfaces may be user-friendly, the underlying algorithmic complexity and unpredictability impose a higher cognitive load on users who must constantly evaluate the validity and trustworthiness of AI-generated outputs.
Key findings regarding trust dimensions include:
Trust in Websites: This dimension, encompassing IT quality, security, and privacy protection, was found to have the strongest influence on users’ perceived usefulness of generative AI. Users prioritize a secure and reliable platform for content creation.
Human-like Trust: This dimension, reflecting users’ perception of AI’s ability, benevolence, and integrity, had the largest overall effect on the acceptance of generative AI. This highlights the importance of emotional connection and perceived social interaction in fostering user adoption.
System-like Trust: While fairness and explainability positively influenced perceived usefulness and acceptance, the impact of accountability and transparency was less significant for general users. The study suggests that technical details or complex transparency information can lead to ‘information overload’ for non-expert users, reducing their experience. Furthermore, the lack of established accountability mechanisms for AI misuse limits trust in this dimension.
Implications for Policy and Development
The study underscores the urgent need for robust regulatory mechanisms and ethical safeguards to build a trustworthy AI ecosystem. The authors recommend that platforms clearly label AI-generated content and provide accessible explanations for AI decisions, especially in high-risk domains like healthcare and finance. Policymakers are urged to clarify responsibility allocation among AI developers, service providers, and end-users, as current legal frameworks often leave the accountability of malicious users ambiguous.
Also Read:
- Public Skepticism Mounts Over AI’s Growing Role in Journalism, Studies Indicate
- Leading Voices Caution Against AI’s Threat to Human Ingenuity and Ethical Standards
Maintaining transparent usage logs and audit trails is also crucial to facilitate prompt accountability and traceability in cases of misuse. By addressing ‘abuse anxiety’ through enhanced transparency, clear accountability, and ethical development, stakeholders can foster greater public trust and ensure the sustainable and responsible evolution of generative AI technologies.
