A cascading failure is a sequence in which the failure of one component changes the inputs or load of the components connected to it, causing them to fail in turn. What defines it is not the first fault but the structure that carries the damage onward.
Example
In an AI agent workflow, a wrong retrieval result at step two is accepted as fact by step three, summarised by step four and acted on by step six. A single early error ends as a confidently wrong final answer. Power grids and supply chains show the same pattern when one failure shifts load onto its neighbours.
See Cascading Failures in AI Agent Systems for the full analysis.