diff --git a/opencti-platform/opencti-graphql/src/database/ai-llm.ts b/opencti-platform/opencti-graphql/src/database/ai-llm.ts
index cd7a506978e5..2b5864637d36 100644
--- a/opencti-platform/opencti-graphql/src/database/ai-llm.ts
+++ b/opencti-platform/opencti-graphql/src/database/ai-llm.ts
@@ -20,6 +20,11 @@ if (AI_ENABLED && AI_TOKEN) {
       client = new Mistral({
         serverURL: isEmptyField(AI_ENDPOINT) ? undefined : AI_ENDPOINT,
         apiKey: AI_TOKEN,
+        debugLogger: {
+          log: (message, args) => logApp.info(`[AI] log ${message}`, { message }),
+          group: (label) => logApp.info(`[AI] group ${label} start.`),
+          groupEnd: () => logApp.info('[AI] group end.'),
+        }
       });
       break;
     case 'openai':
@@ -38,15 +43,17 @@ export const queryMistralAi = async (busId: string | null, question: string, use
     throw UnsupportedError('Incorrect AI configuration', { enabled: AI_ENABLED, type: AI_TYPE, endpoint: AI_ENDPOINT, model: AI_MODEL });
   }
   try {
-    logApp.debug('[AI] Querying MistralAI with prompt', { questionStart: question.substring(0, 100) });
+    logApp.info('[AI] Querying MistralAI with prompt', { questionStart: question });
     const response = await (client as Mistral)?.chat.stream({
       model: AI_MODEL,
       messages: [{ role: 'user', content: question }],
+      safePrompt: undefined,
     });
     let content = '';
     if (response) {
       // eslint-disable-next-line no-restricted-syntax
       for await (const chunk of response) {
+        logApp.info('[AI] Querying MistralAI response chunk', { chunk });
         if (chunk.data.choices[0].delta.content !== undefined) {
           const streamText = chunk.data.choices[0].delta.content;
           content += streamText;