The electronic “smart home” promises endless convenience and security. People will control the temperature of their home from their office. Walmart now lets a delivery worker unlock a house with an app and then stock its refrigerator with food, all monitored by a live camera on the worker’s chest. It all sounds too good to be true. And maybe it is.

Before we plunge headlong into the Brave New World of smart homes, let’s pause and consider potential bugs in the system.

If we’d have done that with Facebook when it started, we might have been able to balance its convenience with the fact that it makes money mostly by creating detailed personal profiles of each of us — and then selling that information. Last July, Facebook was fined some $5 billion by the Federal Trade Commission for violating the privacy of some of its users.

The latest “smart home” controversy involves Cloud Cam, a service that Amazon provides customers so they can integrate the company’s Alexa voice assistant, its Fire TV, and other Amazon products into one package.

An investigation by Bloomberg News reports that “Amazon workers may be watching your Cloud Cam footage.” It reported that humans employed by Amazon each screen up to 150 clips of footage a day from people’s homes in order to improve performance. But some of those clips have included private moments ranging from arguments to sexual activity.

Amazon insists that any footage that is viewed is sent in either by employee testers or voluntarily by customers that allow access to their accounts for troubleshootings. Amazon analysts can finetune the system so it better distinguishes between, say, a cat and a cat burglar.

Amazon has tight security protocols for workers that review Cloud Cam clips. But at no point are customers explicitly told under which conditions their videos might be seen by other people. And since most of their screeners are in low-wage countries such as India and Romania, it is unclear just how seriously Amazon’s protocols are followed in practice. Earlier this year, a separate Bloomberg story noted that human reviewers also listen to Alexa recordings, allowing them to sometimes monitor private conversations.

And accidents happen. Amazon devices depend on microphones listening for a key word and they can be activated by accident when they’re not wanted — thus allowing Amazon employees to listen in.

The current vulnerabilities of the system have prompted internal debate within Amazon. David Limp, vice president of Amazon’s eBook service Kindle, has said: “If I could go back in time, (transparency) would be the thing I would do better. I would have been more transparent about why and when we are using human annotation.”

Few experts believe that big-tech employees listening to some customer interactions mean a new era of Big Brother is upon us. But the glitches that have been uncovered also demonstrate how easy it might be for outsiders to hack the system. Britain’s Daily Mail newspaper reports that in September, British security researcher Mark Barnes was able to turn 2015 and 2016 versions of the Echo Spot security system into a live microphone. “Fraudsters could then use this live audio feed to collect sensitive information from the device,” the Mail concluded.

None of this is an argument against progress or technology. But before we turn ourselves completely over to a network of smart speakers and surveillance cameras, perhaps we should ask the tough and necessary questions about privacy and civil liberties that people in any free society should be worried about.