Article
Understanding prompt injections: a frontier security challenge
OpenAI · openai.com
A plain explanation of prompt injection with examples, and the habits that reduce it (its Atlas examples refer to a browser OpenAI has since retired).
The question it answers
Read with it
- The setting, open to personal accounts since June 2026, that switches off web access, agent mode and connectors. openai.com Article Introducing Lockdown Mode and Elevated Risk labels in ChatGPT OpenAI
- How Claude's browser agent checks each action, its measured attack success rates, and how to go back to approving every step. claude.com Article Claude in Chrome is generally available Anthropic
- Anthropic's own walk-through of a hidden-instruction attack on an email agent, and why 1% success is still real risk. anthropic.com Article Mitigating the risk of prompt injections in browser use Anthropic
- Security researchers show how a web page could steer an AI browser into handing over a user's account details. brave.com Article Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet Brave
- The clearest rule of thumb: an agent with your private data, outside content and a way to send messages can be turned against you. simonwillison.net Article The lethal trifecta for AI agents: private data, untrusted content, and external communication Simon Willison
- How many people already give agents their email (36%) and financial accounts (20%), and the 32% who let AI act without a final... menlovc.com Article 2026: The State of Consumer AI Menlo Ventures
eChai Partner Brands
The partnership →
