Back issues
One a month, whether or not anything happened. Something always happened.
Secure Enough
Issue 3, October 2026
The permission problem
Issue 3: The agent problem
AI agents can act on your behalf. The question nobody asked is what they are allowed to act on.
Secure Enough
Issue 2, September 2026
Maintainer phished by a model, then replaced by a hallucination
Issue 2: The supply chain, hallucinated
Packages that never existed, maintainer outreach written by a model, and the people who still maintain the rest.
Secure Enough
Issue 1, August 2026
Hushly assures users the assistant only shared one customer's credentials, so far
Issue 1: Identity at machine scale
Voice clones that call back, phishing that writes itself, and the passkeys that make persuasion irrelevant.