This article was written with the assistance of AI.
News Factory APP - agentic news to boost your SEO & AEO.
OpenAI rolls out Lockdown Mode to curb prompt‑injection threats
Key Points
- OpenAI launches Lockdown Mode for ChatGPT.
- Feature disables live web browsing, image fetching, deep research, and agent mode.
- Model can still generate images internally.
- Lockdown Mode reduces but does not eliminate prompt‑injection risks.
- Targeted at businesses and users handling confidential data.
- Initially available to self‑serve ChatGPT Business accounts and eligible personal accounts.
- OpenAI emphasizes the setting is a risk‑reduction tool, not a full safeguard.