Credit: VentureBeat made with OpenAI ChatGPT-Images-2.5 Security researchers say they used Anthropic’s newly released Claude Opus 5 to help turn an image-processing vulnerability into an exploit chain ...