Same thing happens in every legacy business process I’ve ever looked at. The rule stays long after the reason for the rule disappeared. Worth checking what you’re still holding onto that you could just verify instead.
A bank in Italy has taken wheels of Parmesan as loan collateral since 1953. Real credit model, cheese ages into real value, the bank lends against it while it matures.
https://t.co/Ck4XLNlLz7
For seventy years the trust mechanism was physical. The bank had to hold the cheese to know it was real. This year they started using blockchain tracking instead, so farmers can keep the wheels at home and still borrow. Lending capacity apparently doubled.
That last line is the entire problem.
Agents don't need less autonomy. They need a name. One person with authority to pause, accountability when it acts.
If this report were written at your company tomorrow, whose name goes on that last line?
INCIDENT AI-2026-014.
Day 1: Renewal agent goes live. Authorized to pull CRM pricing, draft renewals, send anything under $25K without review. Pilot phase, 90 days.
Day 91: Pilot ends. Nobody turns it off. It just keeps running.
Day 140, 9:40am: Sales asks IT. IT points to the workflow. The workflow points to the vendor. The vendor points to the permissions.
Day 141: Legal asks who signed off. Silence.
Root cause: none. All systems performed as designed.
Impact: under review.
Owner: unassigned.
A company adds AI to a workflow, it gets faster, everyone celebrates. Nobody asks why the workflow exists.
You can renovate a room or you can move. Most AI strategies right now are renovations.
Is your company renovating or moving?
Leave any of those blank and you have shipped an identity nobody manages, running at machine speed.
My bet is number 3. Legal points at the business unit, the business unit points at IT, and the agent keeps working.
#AIGovernance#AIAgents#EnterpriseAI
Your next AI agent will read the CRM, send email under a service account, and update records on its own.
You would not hand that to a new hire on day one. right?
Four questions to answer before the agent goes live:
1. Which systems can it reach?
2. Which actions can it take without a human approving?
3. Who owns the outcome when it gets one wrong?
4. Where does its activity land in your audit logs?
Only 10% of organizations are quantum-ready (IBM).
The ones preparing aren't just reducing risk. They're building trust. Trust is the new moat.
What's your first step?
DM to map it out.
@AnthropicAI@IBMQuantum#PostQuantum#AICyber
Anthropic's Claude Mythos just found thousands of zero-day vulnerabilities in every major OS and browser.
One had been sitting in OpenBSD for 27 years.
AI can now outhack all but the most skilled humans. Quantum is the next shoe to drop.
By popular request, GPT-4.1 will be available directly in ChatGPT starting today.
GPT-4.1 is a specialized model that excels at coding tasks & instruction following. Because it’s faster, it’s a great alternative to OpenAI o3 & o4-mini for everyday coding needs.
We’ve rolled back last week's GPT-4o update in ChatGPT because it was overly flattering and agreeable. You now have access to an earlier version with more balanced behavior.
More on what happened, why it matters, and how we’re addressing sycophancy: https://t.co/LOhOU7i7DC
We're excited to announce we’ve launched several improvements to ChatGPT search, and today we’re starting to roll out a better shopping experience.
Search has become one of our most popular & fastest growing features, with over 1 billion web searches just in the past week 🧵