ChatGPT Introduces Elevated Risk Labels Amid Rising Cyber Fraud Concerns

OpenAI has added two new security features to ChatGPT, i.e., Lockdown Mode and Elevated Risk Labels. The aim of these tools is to prevent the stealing of user information by AI. In an era where the use of AI in digital payments, Aadhaar-based services, and online banking is growing exponentially, it is considered an imperative step. This is because, in recent times, cases of cyber fraud and information leaks have been growing continuously.

However, a new cyber threat known as prompt injection has been identified. In this case, a hacker embeds commands in a document. If a user asks the AI to read the document, it ends up revealing sensitive information. For instance, if you are requesting the AI, such as ChatGPT, to read content from a suspicious website and later request it to retrieve data from your system, it ends up revealing sensitive information.

This feature will warn the user in advance that the feature or web-connected tool being used could expose more data. For example, if ChatGPT is connecting to an external third-party website or app, it will clearly indicate the potential risks involved. This allows the user to decide whether to proceed with the chat with the AI.

Lockdown Mode will also prove useful for users. When enabled, ChatGPT will limit external systems, third-party apps, and web connections, reducing the risk of data leakage. This feature can be especially useful for journalists, corporate professionals, government officials, or users dealing with sensitive information. They will be able to have "safe chats" in Lockdown Mode.