Google said there's no threat from a speech recognition feature in its Chrome browser that a developer said could be used to listen in on users.
He created an exploit that could allow a website to continue accessing a computer's microphone after a person thinks they've left a website. Some websites are enabled to use speech recognition, where the website has access to voice commands from a computer's microphone.
"It may seem I have shot myself in the foot by exposing this," Ater wrote. "But I have no doubt that by exposing this, we can ensure that these issues will be resolved soon."
Google acknowledged the problem and had a patch ready by Sept. 24, Ater wrote. The company nominated him for a reward for finding the vulnerabilities, he wrote. Google later decided the issue he found didn't qualify for a bug bounty reward.
But Google never pushed out an update to Chrome. In a statement, Google said it designed the speech recognition feature with security in mind and the feature is in compliance with W3C (World Wide Web Consortium) coding standards.
"We've reinvestigated and still believe there is no immediate threat, since a user must first enable speech recognition for each site that requests it," it said.
Websites enabling the feature ask users for permission to use their microphone first, and Chrome indicates the microphone is active with a red dot in the browser tab.
But Ater found that Chrome remembers if a person granted permission to a site that uses HTTPS, a security feature that encrypts communication between a client and a server. It will allow sites using HTTPS to start listening in the future without asking for permission again.
Ater described a scenario where a website could be configured in a more malicious way to launch a "popunder" window, which is another browser window behind the main one.
If someone navigates off the main page, they may be unaware the popunder window is still active, recording their voice. The popunder window could also be disguised as an advertisement, concealing its true purpose.
"This can be done in a window that you never saw, never interacted with and probably didn't even know was there," Ater wrote.
The spying window could also be programmed to stay dormant until someone says certain, interesting keywords, according to a demonstration video on Ater's site.
The attack doesn't work if permission isn't granted to enable speech recognition.
Send news tips and comments to firstname.lastname@example.org. Follow me on Twitter: @jeremy_kirk
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.