BriefMarket Brief / AI / TechCrunch3 min read
OpenAI introduces Lockdown Mode to reduce ChatGPT prompt-injection data risks
TechCrunch AI reports that OpenAI's Lockdown Mode disables live browsing, web image retrieval, deep research, and agent mode to reduce prompt-injection data-exfiltration risk.
Image source: TechCrunch AI
Key Takeaways
- Even with Lockdown Mode, ChatGPT could be still vulnerable to prompt injections, but the goal is to reduce the likelihood that sensitive data gets shared in the process
- OpenAI announced a new feature that it says will provide additional protection from prompt injection attacks, where malicious chatbot instructions are hidden in webpages and…
- Among other things, Lockdown Mode will disable live web browsing (so you can only access cached content), the retrieval and display of images from the web (you can still generate…
- “Lockdown Mode is not intended for everyone,” OpenAI says
TechCrunch AI reports that OpenAI has introduced Lockdown Mode to reduce the risk that sensitive data is exposed when ChatGPT encounters prompt injection. Prompt injection is the attack pattern in which malicious instructions are hidden inside webpages, files, or other content sources that a model reads.
The mode narrows what ChatGPT can reach outside the conversation. According to the report, Lockdown Mode disables live web browsing, web image retrieval and display, deep research, and agent mode; users can still access cached content and generate images. In plain terms, it limits external browsing and tool-like actions before sensitive work leaves the chat.
OpenAI also says the mode is not a complete defense. Prompt injection can still appear in cached web content or uploaded files and affect the behavior or accuracy of a response, so the feature is meant to reduce data-exfiltration risk rather than remove every attack path.
The report says OpenAI is positioning Lockdown Mode for people and organizations that handle sensitive data, not as a setting every user needs to turn on. The company is rolling it out to self-service ChatGPT Business accounts and eligible individual accounts.
Sources
- OpenAI unveils Lockdown Mode to protect sensitive data from prompt injection attacks
Even with Lockdown Mode, ChatGPT could be still vulnerable to prompt injections, but the goal is to reduce the likelihood that sensitive data gets shared in the process.
- TechCrunch AI AI coverage
TechCrunch AI's current AI coverage page for related reporting and follow-up context.
Tommy
ALTOS LAB product and AI implementation editor, focused on enterprise workflows, generative search and practical decision frameworks.
