NITDA Warns Nigerians Using ChatGPT On New Security Flaws

The National Information Technology Development Agency (NITDA) has issued a fresh cybersecurity alert to Nigerians over vulnerabilities in OpenAI’s GPT-4.0 and GPT-5 series.

NITDA said the vulnerabilities could expose sensitive information and compromise the digital safety of citizens

In a statement on Monday, the agency’s Director of Corporate Affairs and External Relations, Mrs Hadiza Umar, confirmed the discovery of seven critical weaknesses affecting the AI models.

According to the agency, the vulnerabilities allow attackers to manipulate ChatGPT through a method known as indirect prompt injection.

NITDA explained that cybercriminals can hide malicious instructions inside regular online content such as webpages, social media comments, or even shortened URLs. During normal activities such as browsing, summarising text, or using AI-powered search, ChatGPT may unintentionally execute these hidden commands.

Umar further noted that some of the weaknesses make it possible for attackers to bypass safety filters by leveraging trusted domains, while others exploit markdown rendering bugs that can conceal harmful content. Most concerning, she said, is the possibility of memory poisoning, where malicious prompts secretly alter ChatGPT’s behavior and persist across future interactions.

“Although OpenAI has addressed some of these issues, large language models still struggle to fully distinguish genuine user queries from harmful embedded instructions,” she said.

NITDA warned that the implications of the vulnerabilities are significant, ranging from unauthorised system actions and data leakage to manipulated AI responses and long-term behavioural distortion.

Umar further noted that some of the weaknesses make it possible for attackers to bypass safety filters by leveraging trusted domains, while others exploit markdown rendering bugs that can conceal harmful content. Most concerning, she said, is the possibility of memory poisoning, where malicious prompts secretly alter ChatGPT’s behaviour and persist across future interactions.

The agency urged Nigerians, especially tech-driven businesses, digital creators and young professionals who rely heavily on AI tools for content generation, entertainment reporting, and data processing, to adopt responsible usage practices.

As AI continues to shape industries such as entertainment, sports and technology, NITDA emphasised that digital safety must remain a top priority to prevent exploitation by cybercriminals.