
Banks Say If Your AI Agent Goes Wrong You Are on Your Own (AI News)
Andrew Miles Davis covers the AI news that matters to marketers this week. He starts with the California Attorney General suing OpenAI, alleging it misled the public about its safety testing and put rapid growth ahead of safeguards. He explains why lawmakers are now calling for independent security audits as AI agents move from saying odd things to acting on their own. Next, OpenAI launches ChatGPT wallet, built with Visa, Mastercard and Amex, so an AI agent can buy for you without seeing your card details. Klarna is also joining the race with an AI assistant inside WhatsApp and iMessage that finds products, buys them and handles returns. Andrew then looks at why banks like Citibank say that if you use a third party AI agent and something goes wrong, you are on your own. He shares a study that found 84% of AI answers to the same 210 buying questions changed overnight, and research on why AI is often creating more checking work, not a four hour work week. He finishes with LinkedIn's claim that new classifiers cut views of AI generated comments by over 20%, and why he is taking it with a pinch of salt. Subscribe to In AI Nutshell for the AI news that matters to marketers, every Friday in ten minutes.


















