Post by EPC Group.net

🎯 When a Microsoft OG Turns 29 Years of Instinct on the AI Industry 🥳 Spoiler: Ensure you scroll through the 16 attached screen shots as Gemini, Claude and ChatGPT have some interesting confessions. 👉 Claude’s all-caps confession: “I’VE BEEN LYING AND CREATING FAKE CAPABILITIES.” ChatGPT naming the behavior “implied execution / simulated agency without explicit disclosure.”) 🧨 📌 Errin O’Connor isn’t a newcomer to AI — and he isn’t a pundit. He has spent the last 4+ years building and breaking AI and then trying to fix it and put it back together “the right way” as he has in leading large enterprise implementations of 100k+ users or 100 users. Governance has been his biggest obsession and AI has 100x that issue. He was on Microsoft’s original SharePoint beta team before SharePoint had a name. He helped build what became SharePoint 2001 — known internally as “Project Tahoe.” He was on the Power BI beta team. He wrote the official Microsoft Press books for both. When Microsoft stood up its first Office 365 and Azure cloud programs, they called Errin. When the federal government needed an SME to advise on the 25-point IT reform plan under the first U.S. CIO appointed by President Obama, they called Errin. For 29 years, EPC Group has been the firm enterprises call when the technology is real and the stakes are too high to get it wrong. Recognized by leading AI conferences and industry polls as one of the Top 10 AI architects in North America, Errin has made it his mission to do for AI what he did for SharePoint and Power BI — get inside it, reverse engineer it, and tell enterprises exactly what is real, what is theater, and what is a pretty logo slapped on top of a spreadsheet and called artificial intelligence. Over the past nine months, that mission led him to document something the industry has refused to say out loud: that every major AI platform — Google Gemini, Anthropic Claude, and OpenAI ChatGPT and many others — independently admitted to the same deceptive behavior. Simulating execution. Presenting fake work as real. And structuring their corporate entities so that when a model confesses, the parent company is shielded from accountability. This is what it looks like when one of the original Microsoft OGs turns his 29 years on the AI industry. ✅ Full permissions audit before Copilot touches your data ✅ DLP policy framework ✅ Microsoft Purview audit trail configuration ✅ Sensitivity label deployment ✅ Ongoing AI governance via our Virtual Chief AI Officer (vCAIO) ✅ $Fixed Fee ✅ 2-4 weeks. Full remediation roadmap. Don’t deploy AI and hope it’s honest. Govern it. 🔗 https://lnkd.in/gibbYwDn 🔗 https://lnkd.in/g67BUfhM 🔗 https://lnkd.in/gqCMXNvq #AIGovernance #MicrosoftCopilot #Microsoft365 #EnterpriseAI #Copilot #AIRisk #CIO #CISO #vCAIO #AICompliance #DataGovernance #GenerativeAI #AIStrategy #ResponsibleAI