Article
Mitigating the risk of prompt injections in browser use
Anthropic · anthropic.com
Anthropic's research note on how it trains and screens a browsing agent against hidden instructions.
It answers 2 questions
Read with it
- The clearest explanation of when an agent can be tricked into leaking your data, and the one combination to avoid. simonwillison.net Article The lethal trifecta for AI agents: private data, untrusted content, and external communication Simon Willison
- A worked demonstration of a hidden Reddit comment steering an AI browser into a signed-in email account. brave.com Article Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet Brave (Artem Chaikin and Shivan Kaul Sahib)
- What Claude in Chrome can do on a page, why it now acts without approving every step, and how its checks are tested. claude.com Article Claude in Chrome is generally available Anthropic
- Survey data on how many people have tried agents, which accounts they connect and how much agent users spend. menlovc.com Article 2026: The State of Consumer AI Menlo Ventures
- Practical advice such as a separate browser profile for agent work, plus which sites Claude will not open. support.claude.com Guide Use Claude in Chrome safely Claude Help Center
- OpenAI's plain-language explainer of prompt injection, likened to phishing, and the layers it uses against it. openai.com Article Understanding prompt injections OpenAI
eChai Partner Brands
The partnership →
