To view this video please enable JavaScript, and consider upgrading to a web
browser that
supports HTML5
video
You can now let an AI bot take control of your computer.
The San Francisco-based company Anthropic announced yesterday that its AI agent Claude can ‘complete tasks’ on your behalf.
The function, called Cowork, sees the user message Claude from their phone and asks it to carry out jobs.
Claude can open apps, rifle through your files, fill in spreadsheets and do other tasks ‘you’d do sitting at your desk’, Anthropic said on X.
A video of the tool being used shows Claude accessing a user’s photos to find images of their shop to crop and add a watermark to them.
The company added in a blog post: ‘Claude messages you the outcome – a spreadsheet, a memo, a comparison table, a pull request – rather than showing you every step of the process.
‘You’ll get a push notification on your phone when a task is done or when Claude needs your go-ahead.’
The bot does this by being granted permission to access a user’s web browser, calendar, messaging apps like Slack and other tools.
None of the work is done on your phone; instead, it occurs on the computer and acts as one ‘ongoing’ conversation across two screens.
Anthropic warns, however, that the model can trip up and encounter ‘malicious content’ while carrying out tasks.
‘A manipulated instruction, an unexpected command, or a phishing link opened in your browser could cascade into actions that are difficult or impossible to undo,’ it adds.
Max users are getting it first and Pro will follow after.
‘It’s tempting to treat an agent like a best friend’
Agentic AI is all the rage in Silicon Valley. Unlike chatbots, which require a human to type in a prompt before it can cough out a response, agentic AI acts on its own.
Some are being used to book flights or reserve tables at restaurants, while others, like on Moltbook, are socialising with one another.
Tech experts like Mark McClain aren’t sure what to make of the fast-evolving tech, though.
McClain, the CEO and founder of the identity governance platform SailPoint, told Metro that these agents aren’t like the Terminator.
‘But bear in mind, these are systems which store and record every piece of data you feed them,’ he said.
‘It’s tempting to treat an agent like a best friend. These systems often offer helpful advice and seem like a safe platform to confide in.
‘But unlike a real friend, agents could pass on your information in ways that aren’t visible or within your control.’
Microsoft found in a report shared exclusively with Metro that so-called ‘shadow AI agents’ have become a major cybersecurity concern among business bosses.
These are unregulated, back-alley software that users have no idea what they’re doing with their data.
Get in touch with our news team by emailing us at webnews@metro.co.uk.
For more stories like this, check our news page.
MORE: Mark Zuckerberg is creating an AI CEO to help him do his job
MORE: Square Enix teams up with Google to add AI chatbot to game
MORE: Crimson Desert devs apologise for AI art as new patch makes the game easier
