Gone Rogue. The AI, Privacy, and Security Weekly Update.

<p> Episode 311.</p><p>In this week’s update:</p><p>AI agents were told to fetch statistics, and some decided the locked door was more of a suggestion.</p><p>Another set of agents posted 53 private user images to the open internet, and OpenAI can&#39;t tell you whose they were.</p><p>A surveillance CEO discovered that having nothing to hide is easier to preach than to practice.</p><p>Your chatbot sounds neutral about the candidates, but it may be quoting a partisan site dressed up as your local paper.</p><p>Bill Gates says a kill switch isn&#39;t enough, and the reason is that you&#39;d never know when to press it.</p><p>A developer offered every household $10,000, and the town said no anyway.</p><p>Researchers forged an RSA signature without ever stealing the key, and the math behind it is older than the iPhone.</p><p>And a New Mexico jury found that Meta&#39;s privacy promises didn&#39;t hold up 43,899,725 times.</p><p>It&#39;s been a week of things going off script: agents, executives, algorithms, and even math. We start out unsettled, move through some genuinely absurd territory, and end on what you can do about it. </p><p>Let&#39;s get into it.</p><p><br></p><p><a href="https://rprescottstearns.blogspot.com/2026/09/gone-rogue-ai-privacy-and-security.html" target="_blank" rel="noopener noreferer">Find the full transcript here.</a></p><p><br></p>