Field Notes on Trust and AI

RadiationBox

An independent publication on trust signals, media incentives, and AI-native behavior.

Latest Published Essay

A Namecheap outage finally moved my company email to Google

An email to my financial advisor never reached his mailbox. Nine months later, a Namecheap outage turned that warning into about 18 hours without company email.

continue reading

Also Published

What Anthropic said after the Claude watermark debate grew

My original X Article passed 64.7K views. Anthropic's response added detail to the argument; trust still depends on whether the product and detector live up to those claims.

continue reading

Also Published

Why Claude's invisible watermark hit a nerve

Anthropic plans to place an invisible mark inside text from supported Claude models. The argument that followed reached proofreading, translation, code, schoolwork and the question of who wrote the final page.

continue reading

Also Published

AI financial advice needs someone responsible when it is wrong

The same ChatGPT, Claude or Gemini account used for work and dinner ideas is now answering questions about savings and stocks. From my brokerage experience, that leaves open questions about the customer record, risk tolerance, supervision and who is responsible for the recommendation.

continue reading