ChatGPT Security Flaw Enables Phishing via Web Summaries31. May 2026AI Models, Cybersecurity, OpenAIChatGPT implicitly trusts Markdown links in web summaries, making it exploitable for phishing and prompt injection attacks. Share on:
