AI Security: A Human In The Loop Is Essential. It’s Not Enough.
“Human in the loop” has quickly become one of the most reassuring phrases in the modern AI vocabulary. It suggests prudence, restraint, and—above all—control. If a human must approve the system’s actions, what could go wrong?
Human in the loop, often shortened to HITL, describes any arrangement in which a person reviews or authorizes an
Comparing AI Chatbots, AI Agents and Agentic AI
What are AI agents and how do they differ from AI apps most of us have been using for the past few years?
AI Chatbots like Claude, ChatGPT, and Gemini are built on large language models. They respond to prompts and provide answers. They give us information, but don’t take any actions.
AI Agents are
Run Your AI Apps Locally: Why You Should Consider Moving from the Cloud to a Platform You Own
Most lawyers encounter artificial intelligence through cloud services like Claude, Gemini or ChatGPT. They type a question, upload a document, and receive an answer produced on computers operated by someone else.
This arrangement is convenient. It also means that the lawyer is relying on an outside company to process the information, maintain appropriate security, follow
Agentic AI: The Prognosis Does Not Look Good
An AI agent can be attacked by an email you never open. No click. No download. No opened attachment. Just a “confused deputy,” an agent that has failed to distinguish a prompt from data, a problem known as prompt injection.
Three Prompt Injection Vulnerability Examples
Security researchers recently demonstrated prompte injection mechanisms with
Reducing the Risk of AI Prompt Injections: Selected Resources
Prompt Injection is generally acknowledged as the most serious vulnerability in the deployment of AI apps, and AI agents in particular. The Open Worldwide Application Security Project (OWASP), considered by many the world’s leading authority on web-facing system security risks, lists prompt injection as Number One on their list of the top ten
Prompt Injection: The Door Left Open
Agentic AI Conventional software keeps code (instructions) strictly separate from data (the files being processed). Large language models collapse the distinction. To an agent, both are just natural language. A firm’s internal policy and an incoming email are structurally similar. The model cannot reliably tell a document it is meant to read from an order
Agentic AI: It’s A Whole New Bag of Risks
She begged “Do not do that,” then “STOP OPENCLAW.” Neither worked.
That’s what happened to Summer Yue, Meta’s Director of Alignment at their superintelligence safety lab. By the time she reached her desktop to kill the process manually, the AI agent she’d created had already deleted hundreds of emails. You would expect someone with
Agentic AI: Seductive and Dangerous
The hype machine is working overtime on Agentic AI. Don’t fall for it.
AI chatbots merely respond to prompts. They only give you information. AI agents like Claude Cowork or Openclaw go beyond this. They are built on large language models, but can take action on your behalf.
That sounds great, but there is a