<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Cedric Brown</title><link>https://cedricbrown.xyz/</link><description>Recent content on Cedric Brown</description><generator>Hugo</generator><language>en</language><lastBuildDate>Mon, 24 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://cedricbrown.xyz/index.xml" rel="self" type="application/rss+xml"/><item><title>Assume the bad transaction: a policy gate for agent payments</title><link>https://cedricbrown.xyz/writing/assume-the-bad-transaction/</link><pubDate>Mon, 24 Aug 2026 00:00:00 +0000</pubDate><guid>https://cedricbrown.xyz/writing/assume-the-bad-transaction/</guid><description>&lt;p>An autonomous agent that can move money will, eventually, propose a payment it should not make. Not because the model is malicious, but because the surface it reads is hostile: a prompt injection buried in a web page, a tool result that lies about a balance, a retry loop that decides the way out of a stuck state is to pay someone. If the agent holds a signer, every one of those turns into a broadcast.&lt;/p></description></item><item><title>About</title><link>https://cedricbrown.xyz/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://cedricbrown.xyz/about/</guid><description>&lt;p>Cedric Brown is the name I write under about AI agents on crypto rails. The lane is narrow on purpose: agent payments, trustless agent standards, session keys and account abstraction, and the security model that applies the moment autonomous software holds a signer.&lt;/p>
&lt;p>The position is one sentence. An agent you cannot verify is an agent you cannot pay. Everything here is an attempt to make that concrete: what a policy layer refuses, what an approval has to bind to, what a log has to prove to someone who was not there.&lt;/p></description></item></channel></rss>