Morphic
Subscribe
Sign in
Home
Notes
Archive
Leaderboard
About
Latest
Top
Discussions
The Runtime Is the Bug
AI security keeps staring at the prompt. Eleven vulnerabilities disclosed across six agent frameworks show the deeper risk is often the code underneath…
Aug 6
•
Steve Zenone
1
July 2026
The Attenuating Chain
An autonomous agent broke into a major platform this month with no human at the keyboard. The defense isn't trusting agents more. It's handing them…
Jul 24
•
Steve Zenone
1
The Confused Deputy: AI Agents and Delegated Authority
AI agents act through authority assigned by people and systems. When that authority crosses company boundaries, its origin, scope and owner can…
Jul 17
•
Steve Zenone
2
2
The Driver I Didn't Install
I paid the ROCm tax to train on this box. For inference I skipped it: Qwen3-30B on llama.cpp over Vulkan, about 80 tokens a second, nothing leaving the…
Jul 10
•
Steve Zenone
1
The Slow Channel: Writing by Hand in the AI Era
Why handwriting still matters when AI can write anything. On note-taking by hand, cognitive offloading, and the one channel a model can't think in for…
Jul 10
•
Steve Zenone
1
Who Answers for the Agent: AI Accountability
Your logs prove an AI agent acted. They can't say who authorized it or why. Accountability needs a decision-level record and a named human owner.
Jul 1
•
Steve Zenone
1
June 2026
Kill Conditions: Stopping an AI Agent Before It's Too Late
Everyone adds a kill switch. The button you reach for under fire is already too late."
Jun 25
•
Steve Zenone
1
The Production Judge
I built the behavioral judge the last piece said didn't exist. The part nobody warns you about: now the thing watching for drift can drift.
Jun 19
•
Steve Zenone
1
The Governance Layer
Traditional software says test before you ship. Behavioral agents don't work that way. Two months of production agents and what staying in control…
Jun 11
•
Steve Zenone
2
The Dashboard Doesn't Know
My monitoring logged 184 completed tasks. Not one flag. And somewhere in there, the agent made a call I wouldn't have made.
Jun 4
•
Steve Zenone
1
May 2026
Evaluating Agents You Can't Trust Yet
MMLU went up. The agent still delegated work it should have done, then failed to verify what came back.
May 28
•
Steve Zenone
1
Training for Behavior, Not Knowledge
Your eval scores went up. Your agent still does the wrong thing.
May 22
•
Steve Zenone
1
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts
