AI agents blocked as major websites shut the door

Delta and major banks shut out AI agents

Technology companies including OpenAI, Meta and Google have been encouraging users to embrace AI agents, which can navigate the internet and carry out tasks with limited human involvement. 

However, a growing number of businesses are putting limits on what these tools can do.

For anyone hoping to hand over their travel plans or finances to an automated assistant, the restrictions could prove a significant hurdle.

Delta Air Lines has ruled out bookings made by AI agents, while financial institutions JPMorgan Chase and Charles Schwab have confirmed they are restricting bots from accessing customer accounts, citing security risks.

AI agents face new barriers at Amazon and Yelp

Access to business information is also becoming a point of contention. Yelp requires AI agents to use its paid data licensing programme if they want access to its information.

Amazon has taken an even more aggressive stance against automated shopping. 

The company banned Perplexity's shopping agent and sued the start-up last year, arguing that it had continued operating on Amazon's website despite repeated requests to stop.

According to The Verge, Amazon said it had “repeatedly requested” that Perplexity prevent its bot from accessing the platform.

Businesses weigh convenience against risks of letting AI act alone

The restrictions may extend beyond companies that have publicly explained their policies. 

Some users have reported that AI agents which previously worked on Cars.com and Amtrak no longer function as expected, suggesting those websites may also be limiting automated activity.

The issue is not simply whether AI can perform a task. An agent might be able to find a flight, select a product or navigate a banking website, but it still needs permission to interact with the service and complete the action.

That distinction becomes particularly important when an automated system is dealing with payments, personal information or financial accounts.

Businesses face potential security risks when software is allowed to act without a person overseeing every step.

There have already been concerns about the consequences of AI providing unreliable information in Ireland. 

Irish MEPs contacted the Health Minister after chatbots directed women experiencing crisis pregnancies towards anti-abortion websites, prompting questions about how these systems deliver information on sensitive issues.