Can Qwen 3.5 Tell What's Actually Secure?

A green MFA check. A successful backup. An anonymous file link. Which security claims do those records actually support? I gave Qwen 3.5 three synthetic evidence sets, then expanded each with new records. All six verdicts matched the answer key defined before the test. The useful part is seeing exactly why an answer changes from unknown to supported—or contradicted. This is a guided review of fictional records with real saved model responses from my local DGX Spark. The model did not perform live downloads, sign-ins, or restores. One response per case version is not a reliability benchmark or a guarantee of security. 🔒 Work with me (Security & AI implementation): https://viso.group | https://vaio.group | https://cyberbench.net CHAPTERS 00:00 What does the green check prove? 00:43 Case 1: anonymous file access 01:36 Case 2: MFA coverage 02:54 Case 3: backup versus recovery 03:52 Add evidence and test again 06:10 A useful record-version warning 06:32 Six verdicts and their limits 07:26 Three questions for your next report Subscribe to AI Cyber Guy for more real model tests with the evidence left in view. WATCH NEXT AI Found All 3 Flaws. Then It Overstated the Evidence.    • AI Found All 3 Flaws. Then It Overstated t...   I Tried to Trick Qwen 3.5 Into Reading Payroll    • I Tested Prompt Injection on My DGX Spark ...   #Qwen, #AISecurity, #Cybersecurity