The Invisible Watchers: When AI Surveillance Enters the Workplace

Two Harvard dropouts just launched a product that should make every HR professional, privacy advocate, and worker pay attention. For $249, Halo’s new AI-powered smart glasses promise to give wearers “infinite memory” by continuously recording, transcribing, and analyzing every conversation they have. The glasses then display real-time suggestions on what to say next—like having an AI coach whispering in your ear during job interviews, negotiations, or everyday workplace interactions.

This isn’t science fiction. Halo raised $1 million from respected venture capital firms including Pillar VC, Soma Capital, and Village Global. The product is available for preorder as of August 20, 2025. And it represents something far more significant than a clever gadget: it’s a watershed moment in the ongoing transformation of workplace surveillance from exception to expectation.

The technology is undeniably impressive. The AI listens to conversations, processes context in real-time, and surfaces relevant information—whether that’s answering “What’s 37 to the third power?” or suggesting negotiation tactics during a salary discussion. Co-founder AnhPhu Nguyen told TechCrunch their goal is to “make glasses that make you super intelligent the moment you put them on.”

But intelligence isn’t the same as wisdom. And what looks like a productivity tool on an individual level reveals itself as something far more troubling when we zoom out to consider its implications for workplace culture, power dynamics, and human dignity.

The Surveillance That Doesn’t Announce Itself

#

Here’s what makes Halo particularly concerning from an ethics perspective: unlike Meta’s Ray-Ban smart glasses, which include an indicator light when recording, Halo’s glasses have no external signal to warn others they’re being recorded. This isn’t an oversight—it’s a feature.

The founders argue that Meta’s privacy concerns prevent them from releasing similar technology at scale, creating an opening for startups willing to move faster. “Meta doesn’t have a great reputation for caring about user privacy,” said co-founder Caine Ardayfio in the TechCrunch interview, “and for them to release something that’s always there with you… is just a huge reputational risk for them that they probably won’t take before a startup does it at scale first.”

This reasoning reveals a fundamental misunderstanding of why major tech companies hesitate to deploy certain surveillance capabilities. It’s not primarily about reputation—it’s about the recognition that some technologies, regardless of their utility, cross lines that shouldn’t be crossed without robust safeguards, regulatory frameworks, and societal consensus.

The same founders previously created a facial recognition application for Meta’s smart glasses that could identify and dox strangers in public spaces. That project was explicitly designed to demonstrate the technology’s potential for harm. Now they’re commercializing another form of ambient surveillance, this time focused on capturing every spoken word in someone’s presence.

The Workplace Angle: Where Consent Becomes Complicated

#

While Halo markets these glasses for personal use—suggesting applications like acing job interviews or remembering details from conversations—the workplace implications are impossible to ignore. Once these devices become normalized, they will inevitably enter professional environments. And that’s where the ethical complexity multiplies.

Consider a few scenarios:

The Job Interview: A candidate wears Halo glasses to an interview, gaining real-time coaching on how to answer questions. Is this innovation or deception? The employer has no way to know they’re competing against an AI assistant. The power dynamic of interviews already favors employers—now candidates might fight back with covert AI surveillance. But two wrongs don’t make an ethical framework.

The Performance Review: Your manager wears these glasses during your annual review, recording everything you say to later analyze for “sentiment” and “confidence levels.” You weren’t informed. You didn’t consent. The recording becomes part of your permanent employment record, analyzed by AI systems you can’t audit or challenge.

The Confidential Conversation: You confide in a colleague about workplace issues, personal struggles, or concerns about management. Unknown to you, they’re wearing Halo glasses. That conversation is now recorded, transcribed, and stored—potentially forever, potentially accessible to their AI systems, potentially vulnerable to data breaches.

The Client Meeting: A salesperson wears these glasses to every client interaction, gaining an edge through AI-powered negotiation tactics. Is this competitive advantage or surveillance capitalism? What happens when both parties arrive wearing AI glasses, creating a scenario where the conversation is really between two AI systems using humans as avatars?

These aren’t hypothetical edge cases. They’re the predictable outcomes of normalizing ambient workplace surveillance. And they raise questions that go far beyond privacy to touch on fundamental issues of consent, power, and human dignity.

The Broader Context: Workplace Surveillance Is Already Here

#

Halo’s glasses are just the newest entrant in a rapidly expanding market for employee monitoring technology. The data is sobering:

• Approximately 60% of companies now use some form of employee monitoring software, according to industry analyses
• The market for employee monitoring tools is projected to reach $2.7 billion by 2026
• An estimated 96% of employers track employee computer activity in some capacity—from email monitoring to keystroke logging
• Remote work accelerated the adoption of monitoring tools by approximately 65% since 2020

What was once reserved for high-security environments or specific compliance requirements has become standard practice across industries. Companies deploy AI systems that track productivity metrics, monitor communications, analyze sentiment in emails and Slack messages, and even use webcams to ensure employees are “present” during work hours.

The justification is always the same: productivity, security, accountability. And these aren’t entirely baseless concerns. Employers have legitimate interests in ensuring work is completed, protecting sensitive information, and maintaining professional standards.

But there’s a line between reasonable oversight and invasive surveillance. And we’re not just crossing it—we’re obliterating it with AI systems that can monitor more extensively, analyze more deeply, and operate more continuously than any human supervisor ever could.

The Consent Problem in Power-Imbalanced Settings

#

Let’s address the elephant in the room: consent. Defenders of workplace surveillance technologies often argue that employees “consent” to monitoring as a condition of employment. But this framing ignores the fundamental power imbalance inherent in employment relationships.

When your choice is between accepting surveillance or losing your livelihood, calling it “consent” stretches the term beyond recognition. True consent requires:

1. Full information: Understanding exactly what’s being monitored, how it’s analyzed, who has access, and how it might be used
2. Genuine choice: The ability to decline without facing negative consequences
3. Ongoing control: The power to withdraw consent if circumstances change
4. Equal power: Negotiating from a position where “no” is actually an option

In most employment contexts, workers lack all four elements. They’re presented with monitoring as a fait accompli, with limited information about the systems watching them, no real ability to opt out, and no control over how their data is used once collected.

This is particularly acute for workers in precarious employment situations—gig workers, contractors, those in at-will employment states, or anyone in a tight job market. They’re the least able to resist surveillance and the most vulnerable to its potential harms.

The Psychological Toll of Constant Observation

#

Beyond the practical concerns about data security and consent, we need to grapple with what constant surveillance does to human psychology and workplace culture.

Research on surveillance in workplace settings consistently shows negative impacts on employee wellbeing, trust, and ultimately even productivity—the very thing monitoring is supposed to enhance. The panopticon effect—the change in behavior that comes from knowing you might be watched at any moment—creates a culture of performance rather than genuine engagement.

When employees know (or suspect) that every conversation is recorded, every email analyzed, and every action tracked, several things happen:

Trust erodes: The implicit message of comprehensive surveillance is “we don’t trust you.” That message is received loud and clear, and it’s reciprocated. When employers don’t trust employees, employees stop trusting employers—creating a spiral that undermines the collaborative relationships essential for innovative, high-performing organizations.

Creativity suffers: Innovation requires experimentation, and experimentation means sometimes failing, asking “dumb” questions, or exploring ideas that might not work out. Surveillance creates pressure to look productive rather than be productive—to optimize for metrics rather than outcomes. The result is risk-aversion and conformity.

Mental health declines: The stress of constant observation takes a measurable toll. Studies on workplace monitoring link it to increased anxiety, decreased job satisfaction, and higher rates of burnout. This is particularly true for monitoring that feels invasive or unpredictable—like not knowing if a colleague’s fashionable glasses are actually recording everything you say.

Authenticity disappears: Perhaps most fundamentally, ambient surveillance makes it impossible to be fully human at work. We all need moments of candor, vulnerability, humor, and authentic connection. When every word might be recorded and analyzed, people retreat into carefully managed personas. The workplace becomes theater, and we’re all playing roles rather than bringing our whole selves to our work.

What Makes AI Surveillance Different

#

Some might argue that workplace monitoring isn’t new—managers have always observed employees, evaluated performance, and made decisions based on behavior. What’s different about AI?

Several things make modern AI surveillance qualitatively different from traditional supervision:

Scale: Human managers can observe only what’s in front of them at any given moment. AI systems can monitor everything, everywhere, all the time—every keystroke, every word, every facial expression captured on webcam.

Permanence: Human memory is fallible and selective. AI creates permanent, searchable records that can be analyzed indefinitely, used in ways that weren’t anticipated when collected, or weaponized in employment disputes.

Analysis: AI doesn’t just record—it analyzes, categorizes, and draws conclusions. It can detect patterns invisible to human observation, make predictions about future behavior, and potentially encode biases that humans might catch and correct.

Opacity: Traditional supervision is visible and understandable. You know when your manager is watching you. AI surveillance often operates invisibly, using proprietary algorithms that employees can’t examine or challenge. You don’t know what data is being collected, how it’s being interpreted, or what conclusions are being drawn about you.

Power asymmetry: Perhaps most importantly, AI surveillance concentrates information and therefore power. Employers know everything about employees—their productivity patterns, communication styles, stress levels, and more. Employees know almost nothing about how they’re being evaluated, what algorithms are making decisions about their careers, or how to contest automated judgments.

The Regulatory Gap

#

As of November 2025, the regulatory landscape for workplace surveillance remains woefully inadequate to address these challenges. While some jurisdictions have enacted protections—New York City’s 2023 law requiring employers to notify employees about electronic monitoring, California’s restrictions on employee location tracking, the EU’s GDPR provisions on worker data—these are piecemeal responses to systemic problems.

We lack:

Clear federal standards: In the United States, no comprehensive federal legislation governs workplace surveillance. Regulations vary by state, creating a patchwork of protections that sophisticated employers can navigate around.

Technology-specific rules: Most workplace privacy laws were written before AI-powered surveillance became prevalent. They don’t address the unique challenges of machine learning systems that can infer sensitive information from seemingly innocuous data.

Enforcement mechanisms: Even where protections exist on paper, enforcement is often weak. Workers who suspect surveillance violations face significant barriers to investigation and remedy, particularly if they lack union representation or legal resources.

International coordination: In our globalized economy, surveillance systems often cross borders. Data collected in one jurisdiction might be analyzed or stored in another, creating gaps in protection that are easy to exploit.

The Halo glasses illustrate this regulatory vacuum perfectly. There’s no law preventing their use in most U.S. workplaces. There’s no requirement to disclose when they’re being worn. There’s limited recourse for people recorded without their knowledge. The technology has outpaced our legal frameworks, and we’re left scrambling to catch up.

Finding a Balance: Innovation Without Invasion

#

None of this is to say that all workplace monitoring is inherently unethical or that technology can’t play a role in improving work. The question isn’t whether to monitor, but how to do so in ways that respect human dignity, maintain trust, and serve genuine organizational needs rather than reflexive control.

Some principles for ethical workplace surveillance:

Purpose limitation: Monitor only what’s necessary for specific, legitimate business purposes. Don’t collect data just because you can or because it might be useful someday. Each form of monitoring should have a clear justification tied to actual business needs.

Transparency: Employees should know exactly what’s being monitored, why, how the data is used, who has access, and how long it’s retained. This shouldn’t require reading hundred-page privacy policies—it should be clear, accessible, and readily available.

Proportionality: The level of monitoring should be proportional to the risk being managed. High-security environments might justify more extensive surveillance, but an office full of knowledge workers doesn’t need keystroke logging and webcam monitoring.

Data minimization: Collect the minimum data necessary and retain it only as long as needed. Don’t create vast databases of employee behavior that become security liabilities and tempt mission creep.

Human oversight: Automated monitoring systems should support human judgment, not replace it. Important decisions about employees should never be made solely by algorithms without human review and the opportunity for employees to contest conclusions.

Worker participation: Involve employees in decisions about monitoring systems. What feels acceptable varies across organizations and cultures—give workers a voice in determining where lines should be drawn.

Regular audits: Monitoring systems should themselves be monitored. Regular audits should examine whether systems are working as intended, whether they’re creating unintended consequences, and whether they’re still necessary.

The Choice Before Us

#

The arrival of products like Halo’s AI glasses represents a choice point. We can accept ambient surveillance as an inevitable feature of modern work—a trade-off we make for productivity, security, or competitive advantage. Or we can insist that innovation serve human flourishing rather than undermining it.

The most concerning aspect of the Halo launch isn’t the technology itself—it’s the casual assumption that eroding privacy is an acceptable price for individual advantage. The founders’ pitch that users can “cheat” on job interviews and gain an edge in negotiations reveals a fundamentally transactional view of human interaction. Everything becomes a competition to be won with technological superiority, rather than a relationship to be built with trust and authenticity.

This mindset, scaled across organizations and normalized across society, leads to outcomes nobody actually wants: workplaces where everyone is performing for invisible watchers, conversations that are tactical rather than genuine, and relationships mediated by AI analysis rather than human connection.

We can do better. We’ve built incredible technologies—AI systems that can genuinely augment human capabilities, create new possibilities, and solve problems that seemed intractable. But we’ve done it without adequately wrestling with the ethical implications, particularly in contexts where power imbalances already exist.

The workplace is one such context. Employees need protection—not just from illegal discrimination or unsafe conditions, but from surveillance systems that treat them as resources to be optimized rather than humans to be respected. This requires regulatory action, certainly, but it also requires a shift in how we think about the purpose of work and the relationship between employers and employees.

Moving Forward

#

As AI-powered surveillance becomes more capable and pervasive, we face increasingly urgent questions:

• What privacy rights do workers retain in professional settings?
• How do we ensure meaningful consent in power-imbalanced relationships?
• Who owns the data generated by monitoring systems, and who can access it?
• What transparency requirements should apply to algorithmic evaluation of workers?
• How do we preserve space for authentic human connection in surveilled environments?

These aren’t just philosophical questions—they have practical implications for the daily experience of millions of workers. The choices we make now about acceptable surveillance will shape workplace culture for decades to come.

The Halo AI glasses will likely fail or succeed based on market forces—whether enough people want them, whether the technology works reliably, whether social norms accept covert recording. But the issues they represent won’t disappear regardless of Halo’s fate. The next version will be more capable, less obtrusive, and marketed even more aggressively as essential for competitive advantage.

We need to decide what kind of workplace culture we want to build. Do we want environments where trust is replaced by verification, where authenticity yields to performance, and where human connection is mediated by AI analysis? Or do we want to draw lines that protect the dignity, autonomy, and humanity of workers even as we embrace technologies that can genuinely improve work?

This isn’t a choice between innovation and stagnation. It’s a choice about what kind of innovation we want—innovation that empowers and respects humans, or innovation that monitors and controls them.

The answer should be obvious. Making it reality will require vigilance, advocacy, and a willingness to say no to technologies that cross ethical lines, regardless of their claimed benefits. Because some things—trust, dignity, authentic human connection—can’t be recovered once they’re lost to the panopticon of AI surveillance.

The invisible watchers are here. The question is whether we’ll let them reshape our workplaces, or whether we’ll insist that work remain fundamentally human.

References:

• Franceschi-Bicchierai, Lorenzo and Bellan, Rebecca. “Harvard dropouts to launch ‘always on’ AI smart glasses that listen and record every conversation.” TechCrunch, August 20, 2025. https://techcrunch.com/2025/08/20/harvard-dropouts-to-launch-always-on-ai-smart-glasses-that-listen-and-record-every-conversation/
• Cloudflare Research. “AI startup Perplexity accused of scraping websites that explicitly blocked AI scraping.” TechCrunch, August 4, 2025. https://techcrunch.com/2025/08/04/perplexity-accused-of-scraping-websites-that-explicitly-blocked-ai-scraping/
• Industry analysis of workplace monitoring trends, 2025
• Research on psychological impacts of workplace surveillance from organizational psychology literature
• Legal analysis of workplace privacy regulations across jurisdictions