In a significant development, Anthropic has released an upgraded version of its AI model, Claude 3.5 Sonnet, which can interact with web and desktop apps like a human. This innovation has the potential to transform the workplace, but its limitations are still being ironed out. The new model, dubbed "Computer Use," can click and type like a person, but its success rate in completing tasks is still less than perfect.

The AI agent market, expected to be worth $50 billion by 2030, is gaining traction, with Microsoft and Amazon also investing in similar technologies. A recent report by MIT Technology Review Insights found that 49% of executives believe AI agents will lead to efficiency gains or cost savings.

However, despite the hype, AI agents still have a long way to go. Anthropic's 3.5 Sonnet failed to complete tasks successfully in several tests, and its high failure rate raises questions about its adoption in the workplace.

In other AI news, thousands of creatives have signed a petition against unlicensed use of creative works for AI training, and Meta is expanding its facial recognition tests to combat celebrity scam ads. OpenAI has hired a new chief economist and chief compliance officer, and has begun previewing a dedicated Windows app for its ChatGPT platform.