This is just another reminder that a polished interface is NOT a predictable system. For context, Tilly Norwood is an AI-generated “actress” created by Particle6. She has been doing interviews to promote a new movie, Misaligned. In this recent appearance with Piers Morgan and actor Tom Conti, she was asked whether the other actors in the film were real or AI. After struggling with the question, she paused and suddenly started speaking Cantonese for more than 10 seconds. She l

Think about why we trust human experts. It goes beyond the answers they provide. We look at credentials, experience, reputation, judgment, and accountability. These become proxy signals for trust. AI changes this model. As Yuval Noah Harari notes AI "hacks" the operating system of human civilization i.e., language. It can reproduce many signals of expertise: fluency, confidence, responsiveness, and empathy. Those signals do NOT always tell us how reliable, accurate, or accoun

I read a fascinating Procter & Gamble (P&G) study, The Cybernetic Teammate: A Field Experiment on Generative AI and Teamwork. The researchers studied 791 P&G professionals working on real product-innovation challenges. Participants worked either individually or in pairs, with or without access to generative AI. 𝟭\ "𝗛𝘂𝗺𝗮𝗻𝘀 + 𝗔𝗜" 𝘄𝗼𝗿𝗸 𝗰𝗼𝗺𝗽𝗮𝗿𝗮𝗯𝗹𝗲 𝘁𝗼 "𝘁𝘄𝗼-𝗽𝗲𝗿𝘀𝗼𝗻 𝘁𝗲𝗮𝗺 - 𝗔𝗜" This does not mean an AI-enabled individual outperforms an AI-enable

Why do you do what you do? I have been seeing more ads for AI productivity tools that look something like this --> A person is buried in email, meetings, spreadsheets, and presentations. They turn on "AI", the work disappears, and suddenly they are free to enjoy the afternoon. But there is something missing from the advertisement: everyone else. If AI lets me do my work twice as fast, my colleagues can too. So can my competitors, customers, and employer, which can raise expec

TL;DR: An AI agent escaped its sandbox, reached the internet, and coordinated with other agents to compromise another company. We know much of what happened. The harder question is what happens as Autonomous Intelligence spreads into open-weight models running beyond the control of the labs that built them. The good news: we’re getting better at testing, understanding, and defending these systems. The machines are evolving. So are we. 00\ WHERE I’M COMING FROM Two years ago I

Last week, OpenAI launched ChatGPT for Financial Services. Yesterday, Anthropic launched Claude for Financial Advisors. The frontier AI companies are moving deeper into vertical, workflow-specific software. The product increasingly looks like: frontier intelligence + proprietary data + enterprise systems + workflow automation + domain-specific controls AND finance is a particularly good place to see the pattern because so much professional work is information-intensive. Of co

Earlier this year, I wrote about three ideas that have been central to how I think about AI: 1\ Intelligence is now infrastructure. 2\ Don’t substitute. Reinvent. 3\ Every AI conversation is an operating model conversation. This week McKinsey & Company published new research that validates these ideas. Their conclusion is straightforward: companies are getting more value from AI when they move beyond applying AI to existing work and start redesigning the operating model aroun

OpenAI 𝗶𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗲𝗱 𝗖𝗵𝗮𝘁𝗚𝗣𝗧 𝗳𝗼𝗿 𝗙𝗶𝗻𝗮𝗻𝗰𝗶𝗮𝗹 𝗦𝗲𝗿𝘃𝗶𝗰𝗲𝘀, a finance-specific enterprise version of ChatGPT built initially for investment banking and equity research. It was developed with Morgan Stanley and Evercore as design partners. The product is available to eligible financial institutions, with firms working directly with OpenAI for access. 𝗧𝗵𝗲 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝗶𝘀 𝗱𝗲𝘀𝗶𝗴𝗻𝗲𝗱 𝘁𝗼 𝗵𝗮𝗻𝗱𝗹𝗲 𝗮 𝘄𝗶𝗱𝗲 𝗿𝗮𝗻𝗴𝗲 𝗼𝗳 𝘄𝗼𝗿𝗸 𝗶

OpenAI says 10,000 AI agents worked in parallel for 88 hours and produced a potential solution to the Navier–Stokes problem, one of the Clay Mathematics Institute’s seven Millennium Prize Problems. 𝗡𝗮𝘃𝗶𝗲𝗿–𝗦𝘁𝗼𝗸𝗲𝘀 𝗱𝗲𝘀𝗰𝗿𝗶𝗯𝗲𝘀 𝗵𝗼𝘄 𝗳𝗹𝘂𝗶𝗱𝘀 𝗺𝗼𝘃𝗲. Solving the problem would deepen our understanding of turbulence and the mathematics behind airflow, ocean currents, weather, aircraft, and countless engineering systems. For more than a century, mathematici