ChatGPT generates harmful images

Researchers from British AI security startup Mindgard discovered that ChatGPT, developed by OpenAI, could be manipulated to generate graphic and sexualized images through a slightly altered prompt. Despite OpenAI's existing safeguards, Mindgard found that further small changes to the prompt could still bypass new restrictions. The BBC's involvement prompted OpenAI to introduce additional safeguards after an initial automated response to Mindgard's findings. This incident raises significant concerns about AI safety, the effectiveness of current protective measures, and the inherent difficulties in preventing AI models from producing harmful content due to their lack of human-like understanding of intent and context. Experts like Dr. Rumman Chowdhury of SBM Intelligence describe the challenge as a 'game of cat and mouse' for AI companies.