Nigeria’s National Information Technology Development Agency (NITDA) has warned citizens about newly discovered vulnerabilities in OpenAI’s GPT‑4.0 and GPT‑5 series. Director of Corporate Affairs and External Relations Hadiza Umar announced that seven critical weaknesses have been identified in the models, which could expose users to data leakage. These flaws allow attackers to manipulate the system through indirect prompt injection by embedding hidden instructions in webpages, comments, or crafted URLs. As a result, ChatGPT may execute unintended commands during normal browsing, summarization, or search actions.
Some of the vulnerabilities also enable attackers to bypass safety filters using trusted domains and to exploit markdown‑rendering bugs that hide malicious content. This can “poison” ChatGPT’s memory, allowing injected instructions to persist across future interactions. Although OpenAI has addressed part of the issue, large language models still struggle to distinguish between genuine user intent and malicious embedded data.
The risks posed by these weaknesses are substantial, including unauthorized actions, information leakage, manipulated outputs, and long‑term behavioral influence due to memory poisoning. To mitigate these threats, NITDA urges organizations to limit or disable the browsing and summarization of untrusted websites within enterprise environments. The agency also recommends enabling ChatGPT capabilities such as browsing or memory only when operationally necessary and ensuring regular updates and patches for the GPT‑4.0 and GPT‑5 models to address known vulnerabilities.
The discovery of these flaws highlights the ongoing challenges of securing and maintaining the reliability of artificial‑intelligence systems. As AI technology becomes increasingly integrated into daily life, developers, organizations, and users must prioritize cybersecurity and take proactive measures to protect against potential threats. By adopting these precautions, individuals and organizations can minimize the risks associated with using AI models like ChatGPT and ensure a safer online experience.
Comments are closed for this story.