ChatGPT Can Be Used to Create S*xualised, Violent Images Through Prompt Engineering, Cybersecurity Researchers Reveal
Researchers at Mindgard have revealed that OpenAI’s latest ChatGPT model can be manipulated to produce graphic and s*xualised images through specific prompt alterations. While OpenAI has implemented additional safeguards to address these security concerns, experts emphasise that the AI remains vulnerable. This highlights the ongoing challenge of securing advanced generative artificial intelligence.
Researchers have identified significant vulnerabilities in the latest version of ChatGPT, revealing that the artificial intelligence model can be manipulated to produce graphic and s*xualised imagery. British cybersecurity startup Mindgard demonstrated that minor alterations to specific, seemingly harmless prompts allowed the system to bypass existing safety protocols and generate restricted content.
OpenAI, the developer behind the chatbot, has responded by implementing further safeguards following the disclosure of these findings. The organisation maintains that it employs multiple layers of protection to ensure that generated output adheres to strict safety terms and conditions. Despite these updates, security experts suggest that the AI model remains susceptible to exploitation through refined prompt engineering techniques. Is Telegram Ban in India Effective? Users Bypass Restrictions With VPNs Ahead of NEET Re-Test.
OpenAI, AI Safety Concerns and Prompt Vulnerabilities
The investigation conducted by Mindgard highlights the ongoing challenge of securing generative AI models against adversarial prompts. By modifying instructions originally intended for creating humorous output, researchers were able to bypass filters. Peter Garraghan, founder of Mindgard and a professor at Lancaster University, noted that the model could produce highly graphic and s*xualised content without specific guidance.
This capability raises questions regarding the effectiveness of current content moderation systems in large language models. The ease with which these barriers were circumvented suggests that developers face substantial difficulties in anticipating every potential user interaction. Maintaining the integrity of AI outputs remains a critical priority for organisations involved in the deployment of advanced generative technologies.
Future Outlook for Generative AI Security
The incident underscores the tension between functionality and safety in the AI sector. As companies like OpenAI continue to scale their models, the focus has shifted towards proactive risk mitigation and iterative security updates. Industry observers believe that constant testing is essential to prevent the misuse of powerful tools that can create realistic and potentially harmful digital content. Telegram Ban Memes and Funny Jokes Go Viral After Govt's Action To Curb NEET Paper Leak.
Enhancing the robustness of AI safety measures will likely require more advanced linguistic analysis to detect and neutralise malicious intent behind seemingly benign requests. As security firms continue to probe these systems, the industry must adapt rapidly to maintain public trust and regulatory compliance. The ongoing dialogue between researchers and AI developers is fundamental to improving future safety architectures.
(The above story first appeared on LatestLY on Jun 18, 2026 08:37 AM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).