Microsoft Has Tested AI Agents For Online Shopping Security Risks: Here’s What Happened
Microsoft Has Tested AI Agents For Online Shopping Security Risks: Here’s What Happened
AI agents are taking over the internet and companies like Google, Perplexity and even OpenAI are showing us a glimpse into the future of web browsing. But Microsoft has sought caution over the AI hype and excitement and warned users about the risks posed by these agents. The company has done with the help of the Arizona University to create a Magnetic Marketplace with focus on testing the capabilities and value of the AI agents.
This is an important test because AI agents are supposedly going to help you do online shopping and even have the payment details stored to automate the process. But is too much automation going to prove its bane? That's what the researchers at Microsoft have decided to find out.
AI Agents Put To Test: Do They Succeed?
The first thing we can highlight about AI agents is that you have to feed them prompts to act on a task. And that is likely to make them vulnerable to manipulation, something that Microsoft has also observed during its test. The Marketplace was set up in a way that AI agents were asked to order food and make payments as well.
The agents were also trained to do these tasks using a variety of AI models and many expected them to seamlessly move across tabs to let work happen on the sidelines. However, Microsoft has questioned the overall purpose of having AI agents handle end-to-end tasks rather than just assist with these activities.
We've already heard security experts warn about prompt injections making AI agents go rogue, and the last thing you want is for bad actors to access these agents and steal your confidential payment details or data.
It is still early days and like most tech challenges, the industry will gradually evolve to these risks, but the early signs are worrying and companies need to find the best solutions to fix these loopholes.
Swipe Left For Next Video
The hackers have matured with the advancement of tech tools and AI is very much part of their armoury. They can easily use the AI agents to manipulate your actions and that's where the human connection will become more valuable than ever.
Granted, companies are ditching the workforce in favour of AI but there has to be a balance between the two and Microsoft has just given us more reason to be vary of AI and its full control on systems.
Source: News18
Related Posts: Google Workspace Studio For AI Agents Launched Google rolls out Workspace Studio so users can build custom AI agents OpenAI is taking on Google Translate with new ChatGPT translator Border agents involved in fatal shooting of Alex Pretti placed on leave U.S. immigration agents face backlash after Minneapolis killings Federal agents fire chemicals as protesters try to block car at immigration site outside Chicago OpenAI says Chinese agents used ChatGPT to plan mass surveillance tools Using helicopters and chemical agents Alex Pretti's unruly confrontation with ICE agents goes viral Social Network for AI agents- Moltbook goes viral