Halo X: The Controversial ‘Always-On’ AI Glasses From Harvard Dropouts
Real-time transcription, infinite memory—or a privacy nightmare?
A bold vision for wearable AI
Two former Harvard students, AnhPhu Nguyen and Caine Ardayfio, are launching Halo X, a pair of AI-powered smart glasses that aim to transform how people interact with information. Priced at $249 and available for preorder starting Wednesday, the glasses are marketed as “always-on” assistants that listen to, record, and transcribe every conversation—then deliver real-time prompts and answers through the display.
“Our goal is to make glasses that make you super intelligent the moment you put them on,” Nguyen said. Ardayfio described them more bluntly: “They give you infinite memory.”
How they work
Halo X glasses feature a microphone and discreet display, tethered to a smartphone app for computing power. The system relies on Google’s Gemini for reasoning tasks and Perplexity for internet-sourced answers.
- Real-time transcription: Records conversations and deletes audio after transcribing.
- Instant prompts: Pops up definitions, calculations, or suggested responses.
- AI flexibility: Uses different models for math, reasoning, or web lookups.
For now, the glasses don’t include a camera, though the founders admit they’re exploring the option for future versions.
The privacy dilemma
Unlike Meta’s Ray-Ban smart glasses, which display a recording indicator, Halo X is designed to be covert. “We want it to be discreet, like normal glasses,” Ardayfio said. That means no visible cues for people nearby that they’re being recorded.
This design has triggered alarms among privacy experts. Eva Galperin of the Electronic Frontier Foundation warned:
“Normalizing the use of an always-on recording device eats away at the expectation of privacy we have for our conversations in all kinds of spaces.”
In the U.S., several states require two-party consent for recordings. Halo says it leaves compliance up to the user. “We trust our users to get consent,” Ardayfio stated.
A controversial track record
Nguyen and Ardayfio have already sparked backlash for past projects. While at Harvard, they built I-XRAY, a prototype that added facial recognition to Meta’s Ray-Ban glasses. In tests, the system could pull up strangers’ personal information in seconds—demonstrating just how easily smart glasses could be weaponized for doxxing.
Although they never released the code, the project highlighted the risks of blending AI and wearable surveillance.
Data and security questions
Halo relies on Soniox for transcription, which the founders claim doesn’t store audio. Nguyen also promised end-to-end encryption and eventual SOC 2 compliance, though no timeline was provided. Critics remain skeptical about how secure user data will be, and who might gain access to it.
A mix of promise and peril
Halo X is pitched as a tool for “vibe thinking”—a wearable AI that makes users sharper and more knowledgeable in real time. Early backers, including Pillar VC, Soma Capital, Village Global, and Morningside Venture, have invested $1 million in the project.
Yet privacy advocates warn that the normalization of covert, always-listening devices could erode public trust and create new risks. While the founders envision a future where AI glasses boost human intelligence, society may not yet be ready to accept the trade-off between convenience and privacy.
