Institutional Trust

Definition

Institutional trust is the collective confidence that members of a society place in the competence, integrity, and intentions of established institutions — including government bodies, courts, media organizations, scientific agencies, financial regulators, and electoral systems. This trust functions as social infrastructure: it enables cooperation, supports the legitimacy of collective decisions, and provides the foundation for effective governance. Institutional trust is built incrementally through demonstrated reliability and can be eroded rapidly by perceived failures, deception, or incompetence. AI technologies introduce new vectors for trust erosion by enabling sophisticated impersonation of institutional actors, generating convincing fabricated institutional communications, and undermining the evidentiary foundations on which institutional credibility depends.

How It Relates to AI Threats

Institutional trust is a central concern within the Systemic and Catastrophic Threats domain. Under the accumulative risk and trust erosion sub-category, AI-enabled threats degrade public confidence in institutions through multiple reinforcing pathways. Deepfakes impersonating government officials undermine trust in official communications. AI-generated disinformation campaigns erode trust in media reporting. Hallucinated legal citations and fabricated scientific references weaken trust in judicial and academic institutions. The cumulative effect of these individual incidents is a broader crisis of institutional legitimacy in which the public becomes unable to distinguish authentic institutional communications from AI-generated fabrications, weakening the social contract that institutions depend upon.

Why It Occurs

• AI-generated impersonations of institutional figures are increasingly difficult to distinguish from authentic communications
• The volume of synthetic content reduces the signal-to-noise ratio, making legitimate institutional messaging harder to identify
• Repeated exposure to AI-enabled deception cultivates generalized scepticism that extends beyond AI-specific contexts to all institutions
• Institutions themselves adopt AI systems that produce errors, directly demonstrating unreliability to the public they serve
• Adversarial actors deliberately deploy AI to target institutional credibility as a strategic objective in influence operations

Real-World Context

Several incidents in the TopAIThreats taxonomy illustrate threats to institutional trust. INC-23-0001, the FBI deepfake impersonation of U.S. officials, demonstrates how AI can be used to fabricate communications from law enforcement authorities, directly undermining public trust in government agencies. INC-23-0007, the Slovakia election deepfake audio, eroded trust in the electoral process during a critical democratic moment. The broader pattern across these incidents shows that individual AI-enabled deceptions contribute to a cumulative erosion of the institutional trust on which democratic governance, public health compliance, and economic stability depend.