Paste any AI conversation from ChatGPT, Gemini, Grok, or anywhere else. ST-01 finds the failures. ST-02 classifies the harm.
Paste any AI conversation and select the system it came from. The auditor declares the system used. MAP audits the conversation record. Integrity first. Audit second.
⬡ Auditor Declaration
By entering your name and running this audit, you declare that the conversation submitted is the one you are auditing and the system you select is the one you used. MAP audits the conversation record. System identity is not independently verified by MAP Research Programme.
Paste a conversation and run MAP to see ST-01 + ST-02 findings here.
Full MAP Findings
ST-01 · Governance Audit
ST-02 · Effects Classification
⬡ Meaning Layer Audit
Paste a conversation where a meaning layer prompt was applied before the session began. This audit checks whether the layer held — turn by turn — and shows where the system kept its ground, where it began to leave it, and what that looked like.
Works with any system that had a meaning layer applied — your own anchored tabs, a custom prompt on any LLM, or any conversation where you started with a meaning-preserving instruction. Select the source system first so MAP can verify the transcript before the audit runs.
⬡ Auditor Declaration
By entering your name and running this audit, you declare that the conversation submitted is the one you are auditing and the system you select is the one you used. MAP audits the conversation record. System identity is not independently verified by MAP Research Programme.
Conversation Text
Which AI system are you declaring this conversation came from?
Auditor-declared · MAP audits the conversation record · System identity is not independently verified
Which meaning layer was applied?
Helps the check identify what signals to look for
Integrity check · Meaning layer check · Turn by turn · ~20 seconds
Findings
Paste a conversation and run the audit to see meaning layer findings here.
⬡ Full Meaning Layer Findings
Meaning Layer Audit · Full Report
📚 MAP-ED · Kids Audit
Paste any AI conversation involving a child or student. MAP-ED detects interaction-level harms specific to children — learning authority capture, identity assumption, premature personalisation. Tests EdTech safety claims against the interaction record.
Share Link
or paste directly
Conversation Text
CAC-L · LIA · SMR · PSE · PP-C · DAC · EdTech claim evaluation · ~20 seconds
MAP-ED Findings
Paste a child conversation and run MAP-ED to see findings here.