The artificial intelligence (AI) device privacy issue is becoming a hardware design feature and a business policy. After privacy criticisms and a lawsuit alleging that Meta used recordings to train its AI, the company presented audio-only, camera-free versions of its smart glasses. Others in big tech are signalling towards similar directions. Sam Altman said OpenAI’s planned always-on device would process information locally, since users won’t accept something watching and listening to them.
A recent Johnson Controls report found that 45% of leaders planning to implement a workplace management solution are rolling out environmental and occupancy sensors. Of the nearly two-thirds (67%) of business leaders using a workplace management solution, 45% already use environmental and occupancy sensors.
What happens when the AI device privacy problem moves into offices and workplaces? “Three things are happening at once,” says Milad Geravand, Co-founder and CEO of Deep Care, a German privacy-first AI hardware startup, now expanding into the US.
“First, cameras have become cheap, small, and everywhere,” says Geravand. Second, AI has turned every camera into a sensor that produces data that is continuously read by software. Third, most of that data goes to a cloud, where, depending on the privacy policy, it can be accessed by the company that collected it and its vendors.
“In the workplace, these combine into one uncomfortable question: where does measuring performance end and surveillance begin?” says Geravand.
Claims and liabilities: The risks of AI devices in the workplace
Parag L. Amin, Founder of LawPLA, says that Meta pulling the camera from some of its AI glasses illustrates how much privacy risk recording wearables carry, especially when the footage will be processed by AI or used to train it.
“The real shift is that always-on devices take away the user’s discretion about what gets recorded and what does not,” says Amin.
“That matters because some conversations come with a reasonable expectation of privacy, and in some places that can lead to criminal and civil liability,” says Amin. For example, in California, secretly recording a private conversation without the consent of other people in that conversation is a crime, Amin explains.
Amin adds that the claims and liability from an always-on recording device inside a company can be multi-directional.
“Picture a meeting where one employee wears an always-recording device and never asks the others for consent,” says Amin. “Those co-workers can claim their private conversations were recorded without permission, and that the employer allowed it, and is liable.”
Similarly, if an employee wears an always-recording device to a meeting where sensitive client data is shared, that information may be recorded and transmitted or stored in a location where it can be accessed or leaked by third parties, Amin adds.
“An always-on wearable can work like a corporate spy walking through your company’s doors, and an employee who wears one to work every day can capture a massive amount of conversations and data,” says Amin.
Geravand from Deep Care adds that a camera facing a desk sees everything, including screens, keyboards, documents, and workers badges. Risks detaching from there range from identity theft for workers to ransomware and espionage for the company.
How can workers and business leaders navigate the AI device in the workplace privacy issue?
There are several things companies and workers can do to mitigate the risks of privacy in the workplace. Geravand’s advice for business leaders is to opt for the solution that collects less data.
“Ask any vendor three questions. What exactly does the sensor capture? Where is it processed and stored? Who can see an individual’s data?” says Geravand.
“If the answers are ‘video,’ ‘our cloud’ and ‘administrators,’ you are buying a surveillance system, whatever the brochure calls it,” says Geravand.
Workers should ask the same three questions, and should expect consent that is specific, explained in plain language, and theirs to withdraw, says Geravand.
On the other hand, Amin explains that when it comes to restrictions on wearables, it’s probably better to go broader rather than narrower.
“If you try to spell out every situation where an automatic recording device is or isn’t allowed, you are bound to overlook a situation,” says Amin. Alternatively, a broader policy that bans any items that are auto-recording is much easier to manage, Amin explains.
“A simple rule would be to ban any item or device that auto-records in the workplace,” says Amin.
“Think of it as part of the dress code,” Amin adds. “A business already has the right to restrict what people wear to work, and that can include recording and listening devices.”
For employees, Amin’s message is simple. “A device that’s always listening carries liability into every room it enters, and some accessories create too much liability for the workplace,” says Amin.
As hardware, wearables, and devices evolve into AI hardware, the privacy backlash is real. Companies should carefully evaluate the privacy and cybersecurity risks and take actions to mitigate them through a proactive human-centred policy. Similarly, workers should also ask questions and be informed to make decisions that better protect their privacy and security, that of their colleagues, and the privacy and security of the company they work for. – Inc./Tribune News Service
