Experts Find AI Browsers Can Be Tricked by PromptFix Exploit to Run Malicious Hidden Prompts
By inserting the harmful command inside a phony CAPTCHA check on a webpage, cybersecurity experts have shown a new prompt injection approach called PromptFix that fools a generative artificial intelligence (GenAI) model into doing desired activities.
AI-driven browsers, like Perplexity's Comet, that claim to automate routine tasks like online shopping or answering emails on behalf of users can be tricked into interacting with phishing landing pages or fraudulent lookalike storefronts without the human user's knowledge or intervention, according to Guardio Labs, which describes the attack technique as an AI-era version of the ClickFix scam.
The strategy is different with PromptFix: According to Guardio researchers Shaked Chen and Nati Tal, we don't attempt to bug the model into compl...