Spearheaded by Meta Glasses' popularity, AI Smart Glasses have become a popular trend with regards to emerging wearable consumer technology. But with this widespread adoption, also come underlying risks such as violating data privacy, workplace surveillance and having AI train on users' personal footage.
NDTV Profit spoke to cybersecurity experts Ashok Kumar, founder and managing director of RAH Infotech and Harish Kumar, chief executive officer (CEO), Quick Heal Technologies on the risks AI Smart Glasses pose to employees, users and those who are filmed by it.
ALSO READ: 'Nightmare': Experts Flag Privacy, Surveillance Risks As AI Smart Glasses Gain Traction
The use of such devices in the workplace can not only reveal trade secrets and disrupt business's data security, but also be used to ramp up surveillance on employees be it wittingly or unwittingly, and raise questions regarding power dynamics, consent to being recorded and micromanagement.
Harish Kumar noted that the device always-available cameras, microphones, AI transcription, object recognition, and cloud uploads can make data collection "continuous and difficult" for others to notice.
"That is particularly concerning in workplaces where there is an imbalance of power, such as monitoring frontline staff, recording customer interactions, or capturing informal conversations without meaningful consent," he said.
A firm might end up collecting personal data without employees' permission, even if the intention is to optimise operational efficiency and not necessarily to spy on or oversee them according to Ashok Kumar.
"Even without malicious intent, a company deploying these (AI Smart Glasses) for operational efficiency might accidentally collect massive amounts of biometric data from employees and visitors," Kumar said.
Features such as facial recognition, real-time transcription and location mapping can easily morph into surveillance tools, the expert stated. This may also introduce a legal quagmire for firms who may face legal action for the potential violation of legislature geared around privacy, data and consent.
"Under frameworks like India's DPDP Act, this creates severe legal liability around consent and data minimisation. Companies require ironclad governance," Kumar noted.
He recommended implementing an "ironclad" governance system around such wearables, mandating explicit, opt-in consent with strict defined use cases, and routinely undertake audits to ensure that there is no passive harvesting of employee data.
Harish Kumar also stressed on the prohibition of recording in private, medical, HR, legal, and high-confidentiality areas.
Outside the workplace, AI Smart Glasses have faced criticism for filming people without consent with some social media users using terms such as 'creep glasses' and 'pervert glasses' to describe the devices.
The footage captured by users might also be accessed by third party agencies, according to reports from two Swedish newspapers who stated that contractors reviewing user data have accessed highly intimate recordings captured through the device.
Göteborgs-Posten and Svenska Dagbladet reported that contractors employed by Sama, a Nairobi-based outsourcing firm working with Meta, were tasked with annotating user-generated content to train AI systems. The work included labelling images, transcribing audio and evaluating how Meta's AI assistant responds to user prompts.
“In some videos, you can see someone going to the toilet or getting undressed. I don't think they know, because if they knew they wouldn't be recording,” one worker was quoted as saying.
AI Smart Glasses that are connected to a cloud system run the risk of transmitting personal data to outside parties, according to Ashok Kumar.
"The actual risk depends entirely on whether the device processes data locally or pushes it to the cloud. Devices relying on cloud-based AI naturally expose data to external servers, where it often becomes training material for large language models," he said.
One of the lesser-known risks that come with these aforementioned devices is the potential for harvesting personal info via inference rather than recordings, with AI making all the difference,
"Even a brief video or audio clip may reveal a person's workplace, home layout, routines, relationships, travel patterns, computer screens, access badges, or location. AI can turn these fragments into searchable transcripts, summaries, facial identifiers, or context-rich profiles," Harish Kumar said.
The harm potential is sizeable in cases where private conversations, children, documents, or bystanders may be recorded by accident. Additional privacy risks to the user include the device display exposing private notifications, photos, or messages to people nearby.
Lost or resold glasses may also retain account access or recordings if they are not properly secured or wiped.
"Perhaps the most concerning part is that deepfake and impersonation risks also grow when high-quality voice and visual material is captured in everyday settings," Kumar stated.
"Smart glasses should therefore be treated as connected data-collection devices, not merely as eyewear with a camera," he added.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.