Transparency Checklist

A transparent AI agent does more than just provide answers—it makes its reasoning, data sources, and decision-making processes clear to users. Transparency builds trust, enables accountability, and helps users understand the strengths and limitations of an AI system.

Does your agent:
disclose the language models and data sources it uses if asked? *
provide references, or links to sources when outputting facts?
explain its non-obvious decisions and actions?
tell the user when its responses may be influenced by training biases?
explain its reasoning when reasoning is used?
clarify and explain its outputs when asked to by a user or administrator?
show its working and provide summaries when using tools?
explain why, when it refuses to answer a question or perform a task?
log outputs, analyse them and report any bias or repeated decision pathways?
report when its response is personalised based on knowledge of the user?
disclose when its output is opinion, speculation, or subjective?
disclose if its sources are likely to be outdated?
make available system or other relevant prompts when asked? *

* Not all transparency is desirable. Be sure evaluate each case with your technical, leadership and legal teams.

Transparency Risk: 100

Unless you've addressed most of the items in the above checklist, your agent is likely to have a short lifespan. Check the boxes to answer the above questions and get an assessment of your agent's sustainability.

AI Agent capabilities are growing rapidly - and so is the need for transparency. Be sure to check this list regularly for new knowledge.