
Anthropic is using Claude Mythos 5 to scan code, rate vulnerabilities, and support partner security tools.
Claude Code can now open, read, click, and type on external web pages inside the development environment, with safeguards for write actions, purchases, account creation, and CAPTCHAs.

Claude Code now includes an integrated browser window that lets Claude open, read, click, and type on web pages directly inside the app. The feature is designed for external resources such as documentation sites, issue trackers, and similar developer workflows. It works like a tab-based browser and can be opened with a keyboard shortcut.
The browser uses the same tools Claude Code already uses to preview local apps, but adds extra safety checks for external websites. Classifiers screen write actions on external sites, and Claude will not buy anything, create accounts, or bypass CAPTCHAs without user consent. The browser also runs on a clean profile with no saved logins.
Organizations can restrict access to external sites through an allowlist or disable the browser tools entirely. The article also notes that users who want Claude to act within their own logged-in sessions should use the Chrome extension instead. The key takeaway for teams is to treat the built-in browser as a controlled external-access tool, not a replacement for every authenticated browser workflow.

Anthropic is using Claude Mythos 5 to scan code, rate vulnerabilities, and support partner security tools.

Anthropic takes most Vercel AI Gateway spend despite a smaller token share.

OpenAI is previewing a privacy-focused safety system as it competes with Anthropic over enterprise AI data protections.

Cursor’s Origin enters code hosting as GitHub faces developer frustration over outages and performance.