8 steps · about 13 s from sending to done. Only what actually happened is listed.

1. Step 1:You asked

   Your message was sent to Alpha, together with the earlier messages in this chat so the answer keeps the thread.

2. Step 2:Safety check took 0.3 s

   Your message was checked for harmful content. Nothing was flagged.

   This runs first, so a harmful request never reaches the AI.

3. Step 3:Searched the web took 2.2 s

   Searched for “EU AI Act deployer obligations public authority” on Oct 8, 2026, and read 2 pages from the open web and Wikipedia.

   The numbers in the answer, like [1], point to these pages so you can check them.

   See 2 sources
4. Step 4:Wrote the answer first words after 3.3 s

   Mistral Large 3, made by Mistral AI, running on Mistral AI, wrote the reply word by word.

   It was told to stick to the sources above.

5. Step 5:Checked the answer

   Each sentence was checked before you saw it. All clear.

6. Step 6:Checked for made-up details

   Looked for invented links, sources or names. Nothing needed fixing.

   AI can sound sure about things that aren’t real — this catches the common cases.

7. Step 7:Saved to your history done after 13 s

   Stored by Current AI in Alpha’s own database. You can delete it any time in Data & privacy.

8. Step 8:Second opinion

   You asked 2 other open AIs to answer too — they showed high agreement.