This impressive AI can take over your computer, which has me a little scared

Key Takeaways

Anthropic has introduced a new capability to its AI model Claude: computer use.
The feature is still experimental and may have errors, but Anthropic aims to improve it rapidly.
Anthropic’s Claude AI is only available to developers currently, but in the future could assist consumers with booking flights and conducting research.

Another day, another advancement in artifical intelligence (AI). This time, though, it involves AI actually acting, rather than just generating.

Anthropic, an AI startup backed by Amazon, recently announced a new update to its Claude AI model that allows it to use a computer. You read that right, it can use a computer. I don’t know whether to be terrified or impressed or both.

Like OpenAI’s ChatGPTand Google’s Gemini, Anthropic has been developing its own AI chatbot called Claude. With the company’s new announcement today, its AI model now has the capability to use a computer to perform tasks such as filling out a form or spreadsheet.

“Available today on the API, developers can direct Claude to use computers the way people do — by looking at a screen, moving a cursor, clicking buttons, and typing text,” Anthropic said in a news release. “Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public beta.”

The company says the new feature is still “experimental” and subject to errors. Anthropic hopes by releasing the feature early it can get feedback from developers, and it expects Claude’s computer use capabilities to “improve rapidly over time.”

AI safeguards can’t keep up with the pace of innovation

A viral fake image of a hurricane “victim” illustrates just how far tech needs to grow to protect against the misuse of AI.

The next stage of AI is taking action

AI is advancing rapidly, going initially from generated photos to now the ability to take action

Anthropic released a demo video featuring Claude’s new AI computer use feature in action. In the demo, Claude is tasked with filling out a vendor request form, and fills out the form using information scattered across various parts of the user’s computer. The video is an interesting window into how AI models like this, and others coming in the future, could soon operate our computers for us.

Speaking to CNBC, Anthorpic said that it hopes in the future, consumers will be able to use Claude to help book flights, schedule appointments, and conduct online research, to name a few.

Anthropic admits that Claude’s ability to use a computer is “imperfect.” The company points out that actions that we humans do on a computer effortlessly, such as “scrolling, dragging, zooming-currently present challenges for Claude.” In terms of safety, Anthropic is taking measures to ensure Claude is used in a safe and secure way, and has taken a “proactive approach to promote its safe deployment” to combat threats such as misinformation or fraud.

An AI computer use feature like this is game-changing, and in time could revolutionize the way humans use AI. Innovation like this is impressive to see, but at the same it’s hard not to worry about how much AI could be doing for us in the future, and its potential effect on jobs.