Over-reliance
Trusting the output because it is well written, not because it is right.
Over-reliance is accepting a confident, fluent output without checking it. The fluency is the trap: text that reads well feels reviewed, so the reader stops doing the review themselves.
In a working team this shows up as nobody reading what the agent wrote before it goes to a customer. The cost of checking is low compared to the cost of a wrong message, but the checking is boring and the output is not.
The countermeasure is procedural rather than technical. Require the agent to cite where each claim came from, ask it to flag what it is unsure about, and make a human sign off on anything that leaves the organisation. Confidence in the output is not evidence about the output.
Related terms
Sycophancy
The model agrees with you because you believe it, not because it is right.
Hallucination
Output that is fluent and wrong. Two different problems hide under one word.
Human in the Loop
A person who reviews, approves, or corrects the agent's work at key points.
Grounding
Tying the answer to something checkable, so it can be verified instead of trusted.
Evaluation
Measuring whether the system does the job, on a fixed set of cases, repeatedly.
Want this working inside your team?
It starts with one conversation about your workflow.