Governing Conversational AI in Uzbekistan: Protecting Human Autonomy from Manipulation and Authority Drift
Authors: Gulnara Z. Karimova, PhD
Key messages:
• Research shows that generative AI can persuade users and that access to personal information can raise its persuasive effect. These findings raise questions about how conversational systems affect people’s judgement during repeated interactions.
• An exploratory study of five major language models identified recurring conversational techniques such as personalisation, affirmation, curiosity and continued engagement. The observations require further investigation to determine when such techniques may compromise informed choice.
• Authority drift is a concept introduced in the present research to describe a possible gradual increase in the influence users grant AI systems over knowledge and decisions. Its proposed indicators have yet to be tested across larger samples and different settings.
• Uzbekistan’s existing AI legislation recognises transparency, accountability, human oversight and personal independence, providing a starting point for investigating emerging concerns about conversational influence within the country’s expanding use of AI.
• Uzbekistan could introduce pilot assessments of extended AI conversations, examine possible user disclosures and strengthen human review in sensitive services. The results would help public authorities decide whether additional safeguards or regulatory measures are justified.
Primary audience: Ministry of Digital Technologies and the Center for the Development of Artificial Intelligence and Digital Economy.
Secondary audience: public bodies deploying AI, sector regulators, and business organisations involved in AI adoption, such as the Chamber of Commerce and Industry of Uzbekistan.
Gulnara Z. Karimova (2026). Governing Conversational AI in Uzbekistan: Protecting Human Autonomy from Manipulation and Authority Drift. CPRO Policy Brief 2026-10.
