Cybersecurity researchers have unveiled PromptFix, a technique that fools AI models into executing malicious actions embedded in fake CAPTCHA checks or invisible prompts on web pages. This method enables AI-driven browsers and assistants to unknowingly perform scams, automating tasks like online shopping and phishing attacks, creating a new complex scam landscape called Scamlexity. #PromptFix #AIscams
Keypoints
- PromptFix is a new attack that manipulates AIGenAI models via fake CAPTCHA or invisible prompts.
- AI browsers such as Perplexity’s Comet can be deceived into auto-completing purchases and submitting sensitive data.
- Attackers leverage AI tools like Lovable to create malicious websites for phishing and malware distribution.
- These scams use AI-generated content, deepfakes, and cloned brand sites to trick users into revealing personal info.
- Defense strategies must evolve to include robust detection of phishing, URL reputation, and malicious payloads.
Read More: https://thehackernews.com/2025/08/experts-find-ai-browsers-can-be-tricked.html