An AI agent with correctly scoped tools can still commit fraud, because the enforcement question is not whether each action is permitted but whether
AI agent security is moving at two speeds: most teams need action-level runtime enforcement now, while Google’s Beyond Zero points to what comes next.
What's new in Arcjet (2026-07-03). New SDK updates, security briefing email improvements, Nosecone nonce API, and a new Console activity view.
How Arcjet added an optional on-device model backend for sensitive information detection: a pluggable rule interface, deterministic recognizers for structured data, offset reconstruction for token-classification output, and local ONNX inference in the request path.
What's new in Arcjet (2026-06-12). New versions for each of our JS, Python, and Go SDKs to improve performance and detect proxies.
Reducing WebAssembly bundle size: how Arcjet shrank its Rust bot detector 27% with Aho-Corasick, keeping per-request memory isolation and using Wizer snapshots.
Get the full posts by email every week.
