Companies scramble to update rules for AI agents and generative AI

In short: Many companies are trying to catch up on internal rules for using fast-changing AI, especially AI agents that can take actions on their own.

What's going on

Companies are adopting generative AI, which creates text, images, or code, and “agentic” AI, which can plan tasks and carry them out (like a digital assistant that does not just suggest steps, but actually takes them). The Financial Times reports that leaders are worried about ethical and reputation risks, while the technology keeps moving quickly.

IBM’s UK and Ireland chief executive Leon Butler said companies are playing catch-up on governance, meaning the policies and oversight that set limits and assign responsibility. An IBM survey found that 80 percent of executives in Europe, the Middle East, and Africa said an AI-related incident reached their board or top executives in the past year. Only 40 percent said they changed their AI governance policies afterward, and just 37 percent said their governance is designed to scale as AI agents are deployed more widely.

Some large companies are adding more specific guardrails. Ikea says it has seven core principles, and it sets boundaries that can go beyond the law, including limits on AI-driven surveillance and automated decisions with major impact unless a human is involved. JLL uses a risk grading system, and it would treat an AI agent that can approve vendor payments as high risk and require review.

What to watch

A separate Wharton study found US managers do not agree on who is responsible when AI makes mistakes, whether it is a risk team, a manager, employees, or the vendor. The FT also notes a growing skills gap, with demand for AI risk and governance skills rising faster than the supply of workers who have them. Expect more companies to formalize training and review processes so AI use is easier to track, like keeping a clear checklist for when new tools are allowed into the business.

Source: Financial Times