Artificial intelligence is reshaping the information environment in which democratic systems operate, accelerating the erosion of public trust through disinformation, deepfakes, and automated influence operations. This article examines how AI-driven manipulation undermines democratic safeguards by destabilising evidentiary standards, weakening institutional accountability, and amplifying uncertainty within public discourse. By integrating perspectives from cybersecurity, digital forensics, and AI governance, the analysis shows how synthetic media and algorithmic amplification transform disinformation from a communication problem into a systemic governance challenge. Deepfakes erode the shared foundations of proof, while AI-enabled content distribution exploits platform vulnerabilities and compromised data ecosystems, blurring the boundary between cyber operations and information warfare. The article argues that effective AI governance must function as a form of democratic defence, embedding forensic traceability, evidentiary validation, and institutional oversight into the lifecycle of AI systems that influence public perception. Without these safeguards, AI risks becoming an unaccountable actor in democratic processes, accelerating legitimacy crises and institutional fragility. The study concludes that restoring public trust requires governance models capable of translating technical detection into credible, transparent, and contestable decision-making.