NITDA Warns Nigerians Of Data Risks In ChatGPT GPT-4 & GPT-5 Models
By TOSI ORE
THE National Information Technology Development Agency (NITDA) has issued a warning to Nigerians about potential data security risks in OpenAI’s ChatGPT GPT-4 and GPT-5 models. The alert, delivered by NITDA’s Director of Corporate Affairs and External Relations, Mrs. Hadiza Umar, in Abuja on Monday, highlighted newly discovered vulnerabilities that could expose users to data leakage.
According to Umar, the agency identified seven critical weaknesses in the AI models, which could allow attackers to manipulate ChatGPT through indirect prompt injections. These attacks can involve embedding hidden instructions in webpages, online comments, or crafted URLs, prompting ChatGPT to execute unintended commands during browsing, summarisation, or search activities.
She explained that some flaws enable attackers to bypass safety filters using trusted domains, exploit markdown rendering bugs to conceal malicious content, and even poison ChatGPT’s memory so that injected instructions persist across future interactions.
Despite OpenAI addressing some issues, Umar warned that large language models still struggle to differentiate genuine user intent from maliciously embedded data, posing risks of unauthorized actions, manipulated outputs, and long-term behavioral influence.
