Community

Submit a Failure Mode

Submit a real AI failure mode you have seen in an LLM, agent, RAG system, or enterprise AI workflow. Selected submissions may be featured in the Community Failure Modes registry, and our team may provide free feedback on detection signals, eval design, monitoring strategy, and mitigation options.

Failure mode title *

Short description *What happened, in plain language.

System type *Select all that apply.

Where it appeared *Context such as support bot, internal copilot, research agent, etc.

Observed symptoms *What signals indicated failure.

Impact / severity *

Severity explanation *

How it was detectedEval, red team, monitoring, user report, trace review, etc.

How it was fixed or mitigatedAllow unknown or not yet fixed.

Optional public linkLinkedIn, Medium, GitHub, blog, webpage.

Link label

Contributor name *

Company / affiliation

Contributor email *Required for review and feedback follow-up. Not displayed publicly.

Required confirmationsI give FailureModes.ai permission to publish this submission, edited for clarity, in the Community Failure Modes registry.I give FailureModes.ai permission to contact me for clarification or feedback.I confirm this submission does not include confidential customer data, private logs, credentials, personal information, proprietary prompts, or unauthorized vulnerability details.

Submissions are reviewed manually. Publication and feedback are not guaranteed.