OpenAI is prioritizing safety and security on the platform with its latest rollout of the "Lockdown Mode," which is a new setting within ChatGPT that offers more protection for users in the face of attacks.

There are massive threats that lurk online, and one of them is the infamous "prompt injection attack" that tricks chatbots and models into revealing information threat actors want to extract, steal, and more.

OpenAI Rolls Out Lockdown Mode to ChatGPT Security Feature

Sam Altman's AI company has recently released a support document which confirms the rollout of its new Lockdown Mode feature that aims to help protect users from threats online.

OpenAI has claimed that Lockdown Mode is an "optional advanced security setting that limits many tools and capabilities in OpenAI products that can connect to the web or external services." In simple terms, this means that OpenAI will disconnect its ChatGPT platform, apps, models, and more from the internet and other external services to avoid attacks of any kind.

According to OpenAI, the Lockdown Mode feature is only available for users who have created accounts on the platform. Accounts have to be logged in in order to take advantage of the feature.

However, OpenAI said that it is still in the process of rolling out Lockdown Mode to different types of accounts, including eligible personal accounts, Free, Go, Plus, Pro, and self-serve ChatGPT Business accounts.

To access Lockdown Mode, users need to go to Settings > Safety and security > Advanced security > and select Lockdown mode.

Lockdown Mode to Protect Against Prompt Injection, Attacks

Engadget said that OpenAI rolled out Lockdown Mode for people and organizations who want stricter protection from potential data exfiltration or extraction risks, especially those that stem from prompt injection attacks.

Prompt injection attacks are developed by bad actors following the rise of AI chatbots in the tech industry as a way to trick platforms like ChatGPT. The attack focuses on using specific hidden languages that, most of the time, could not be easily distinguishable to people, even to trained eyes.

The attack centers on adding hidden language that only chatbots could understand or see, and within the language are instructions that can manipulate these platforms to do the attacker's bidding.

This includes divulging personal and sensitive information like addresses, credit card information, login credentials, and more.