Story
July 14, 2026
OpenAI Introduces 'Lockdown Mode' for ChatGPT to Mitigate Data Theft
OpenAI has launched a new security feature for ChatGPT called "Lockdown Mode," which is designed to protect against prompt injection attacks that can lead to data theft. The mode works by disabling risky features such as live web browsing, agent mode, and file downloads.
OpenAI has introduced a new “Lockdown Mode” for ChatGPT, aiming to curb data theft from prompt injection attacks while openly acknowledging that the fix is only partial.
Early concern over prompt injection
Security researchers and AI companies have grown increasingly worried about prompt injection attacks, where hidden instructions in webpages or uploaded files trick models into leaking data or calling external services. OpenAI has described prompt injection as a “frontier” problem affecting all large language models, because they “cannot reliably separate data from instructions.”1
OpenAI announces Lockdown Mode
On June 6, OpenAI detailed Lockdown Mode as a setting that disables live web browsing, image retrieval from the web, deep research, and agent mode to reduce the channels attackers could use to exfiltrate sensitive data.2 Users in this mode can access only cached content instead of making new live requests to external sites.2
The company emphasized that Lockdown Mode is aimed at “people and organizations that handle sensitive data and want stricter protection from data exfiltration risks related to prompt injection,” and that it is “not intended for everyone.”2
Rollout across ChatGPT plans
By June 7, coverage noted that OpenAI had begun rolling out Lockdown Mode to logged‑in users across Free, Go, Plus, Pro, and self‑serve ChatGPT Business plans.1 The feature blocks live browsing, agent mode, deep research, image retrieval, Canvas networking, and file downloads, cutting off “the outbound pathways an attacker would use to exfiltrate the data.”1
Benefits and trade‑offs
Both reports stress that Lockdown Mode “does not guarantee that data exfiltration cannot happen,” with risks remaining via enabled apps or novel attack techniques.12 The protection comes with a significant usability trade‑off: ChatGPT loses much of what makes its agent and research features powerful, prompting OpenAI to frame the mode as an optional safeguard rather than a default for all users.1
[1] The Next Web
OpenAI adds Lockdown Mode to ChatGPT to block data theft from prompt injection attacks.
[2] TechCrunch
OpenAI Unveils Lockdown Mode to Protect Sensitive Data from Prompt Injection Attacks.