Warning Issued on Web Programming Interfaces
Tools that connect websites can also open up new security vulnerabilities, experts say.
The rapid growth of Web applications has been fueled in part by application programming interfaces (APIs)–software specifications that allow sites and services to connect and interact with one another. But at the DEFCON hacking conference in Las Vegas last weekend, researchers revealed ways to exploit APIs to attack different sites and services.
APIs have been behind the meteoric rise of many key social sites. The social-networking site Facebook, for example, won huge gains in popularity and attention after opening its site to applications written by outside developers using its API.
The API of the microblogging media darling, Twitter, is also credited with partly driving its popularity. John Musser, the founder of Programmable Web, a website for users of mashups and APIs, says that the traffic that comes into Twitter through APIs–for example, from desktop clients–is four to eight times greater than the traffic that comes through its website. “The API has been crucial to the success of that startup,” he says.
But researchers Nathan Hamiel of Hexagon Security Group and Shawn Moyer of Agura Digital Security say that APIs could also be exploited by hackers. They note that several APIs are often stacked on top of each other. For example, an API might be used by the developers of other websites who, in turn, publish APIs of their own. “There could be security problems at the different layers when this sort of stacking happens,” Hamiel says.
Hamiel also notes that APIs can open sites to new kinds of threat. For example, he points to APIs for building applications that work across multiple websites. These tools may allow developers to pull in content from third-party websites, but Hamiel says that this also opens up possibilities for attacks.
During his presentation Hamiel showed that an attacker might be able to use an API in unintended ways to gain access to parts of a website that shouldn’t be visible to the public. “Whenever you add functionality, you increase your attack surface,” Hamiel says, noting that what makes an API powerful is often the same as what makes it risky.
Programmable Web’s Musser says that many of the security risks introduced by an API are similar to those found in desktop computers. In both cases, he says, security vulnerabilities exist wherever there is an access point that an attacker might abuse. Any site that builds its API on top of another site’s API is relying on someone else’s security, and it’s not easy to look into what has been built to see how well it has been handled, Musser says. “Part of the fundamental issue is just how new the technology is,” he adds.
Jeremiah Grossman, founder and chief technology officer for WhiteHat Security, says that sites that publish APIs can find it hard to discover security flaws in them. He notes that often it’s difficult to tell how a third-party site is using an API, and if that site has been compromised by an attacker.
APIs are also harder to test than traditional websites, Grossman says. Though software tools have been developed that can analyze a site’s underlying code to pinpoint potential vulnerabilities, those tools won’t work for testing APIs. “It’s a lot more manual with a lot less automation, and it means, at the end of the day for the business, more expense,” he says.
But while experts agree that there’s no easy fix for the risks introduced by APIs, they also say the technology isn’t going away. “Websites are becoming Web services, and that trend isn’t going to stop,” Musser says.
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today