A chatbot can generate partisan political content, but partisan does not automatically mean disinformation. Disinformation is false or misleading information deliberately created or distributed to deceive or manipulate. In a specific case, deciding whether content is disinformation—or even misinformation—can be contested. A biased viewpoint may still be permitted political expression. section 5.5
Disinformation poses different challenges for social-media platforms and chatbots. Platforms set content policies, but drawing a line between misleading claims and permitted viewpoints can provoke disagreement. Chatbot providers face a further difficulty: because generated answers are probabilistic, they may struggle to explain a particular response or change how the system responds next time. section 5.5
Governance can address how disinformation spreads as well as what individual posts say. The European Union’s Code of Practice on Disinformation emphasizes access to platform data for researchers and fact-checkers and transparency in advertising, rather than prescribing specific content standards. Under the EU Digital Services Act, large platforms face risk assessments, mitigation duties, and audits; an AI recommendation system that contributes to disinformation’s spread could expose a company to fines. section 5.5 section 10.30
