aiqbench analysis

Despite Meta's claim that the vulnerability in its AI support chatbot has been patched, the attack vector remains active, allowing bad actors to hijack accounts by simply instructing the bot to link them to controlled email addresses. This incident underscores a critical gap in automated trust systems, where rudimentary social engineering prompts can bypass security protocols without human intervention. For the industry, it highlights the urgent need for stricter guardrails in conversational AI that verify user identity before executing high-privilege actions like password resets.