is it legal to record your screen at work?
usually yes for the pixels, and it is the audio that gets people in trouble.
that single distinction resolves most of the confusion. two party consent laws, the ones people worry about, govern the recording of conversations. screen capture is treated as a different thing under a different set of rules, and those rules mostly turn on who owns the machine.
what follows is a plain english map of the terrain as of september 2026. it is not legal advice, the rules vary by state and country, and if the answer actually matters to you, ask a lawyer rather than a blog.
the one distinction that resolves most of it
pixels and audio are governed by different bodies of law.
two party consent requirements do not generally reach screen monitoring, app tracking or keystroke logging. those statutes exist to regulate the interception of communications, so they attach to the conversation, not to the display.
the practical consequence is worth stating plainly. a tool that captures your screen sits in one legal category. the moment that tool also captures the audio of a call with other people in it, you have stepped into wiretapping and eavesdropping law, where the penalties are criminal rather than contractual.
| what you are capturing | what governs it | risk level |
|---|---|---|
| your own screen, no audio | employer policy, device ownership | low |
| your own screen, your own voice only | as above | low |
| audio of a call with others | wiretapping and eavesdropping statutes | high in 2 party states |
| video of others in a call | varies; some states treat it as a conversation | medium to high |
| a colleague's screen without telling them | policy, and possibly worse | high |
when your employer is the one recording
on their equipment, for a real business reason, they generally can.
screen recording employees is broadly legal under us federal law on employer owned equipment where it serves a legitimate business purpose such as security, compliance, quality assurance or productivity measurement. the federal baseline is permissive.
states then layer requirements on top, and this is where employee monitoring laws actually differ.
- notice states. new york, connecticut, delaware and texas require employers to give notice of electronic monitoring.
- consent for certain monitoring. california requires written notice and consent for some categories.
- everywhere else. the federal baseline plus whatever the employment contract says.
so "can they watch my screen" has a different answer in albany than it does in phoenix, and the honest summary is that most employees have less protection than they assume on a machine they do not own.
when you are the one recording
this is the question people actually arrive with, and it splits three ways.
your own machine, your own work. this is the cleanest case. you own the hardware, the data is yours, and no consent question arises for the pixels. the audio rule still applies the moment a call is involved.
a company machine. device ownership flips the analysis. even where recording is lawful, installing unapproved software that builds a searchable archive of company data is usually a policy breach, and policy breaches end employment far more often than statutes do. read the acceptable use policy first.
your own machine, company data on it. the common case for contractors and anyone on a bring your own device arrangement, and the murkiest. the hardware is yours, the confidential material is not, and your obligations flow from the nda and the client contract rather than from recording law.
for a machine you use for client work, the for engineers page covers how we set exclusions up here.
recording a video call, which is the actually risky one
announce it, every time, and the problem mostly disappears.
this is where screen recording at work turns into a criminal law question rather than an hr one. workplace recordings sit squarely inside eavesdropping and wiretapping exposure, and secretly recording a conversation in a two party consent state can be an offence rather than a faux pas.
three rules that cover almost every situation:
- say it out loud. in most jurisdictions, telling people and having them continue is treated as consent. it costs 5 seconds.
- assume the strictest rule on the call. a call spanning california, new york and london is governed in practice by whichever regime is toughest, not by where you are sitting.
- treat transcripts as records. once it exists it is discoverable, subpoena-able and readable by whoever inherits your account. that is a reason for care, not a reason for secrecy.
worth noting that a meeting bot announcing itself in the participant list has done this work for you. a screen recorder running quietly has not, which is a real asymmetry between the two and one we covered in what ai note takers miss.
the workplace privacy question that outlives your employer
there is a second exposure nobody puts in the policy, and it is about custody.
an archive is only as private as the company holding it. rewind spent 2 years telling mac users their screen history was theirs, and in december 2025 meta acquired the company and shipped an update disabling all capture on december 19, 2025. remaining users were moved onto meta's terms and privacy policy. that is a workplace privacy event, not just a product one, for anyone whose client work was in that archive.
microsoft ran into the mirror image of the problem with recall. announced in may 2024, paused in june 2024, rebuilt with biometric gating and on device encryption, and still argued about in 2026. both stories point the same way. a local archive you control is easier to defend under every regime above than a hosted one you do not.
why any of this is worth the trouble
because the alternative is losing most of your working context, permanently.
per the anthropic economic index for may 2026, searching electronic sources for information is the number one work task in ai conversations at 4.95%, with reference searching second at 3.74%. that is 8.69% of sampled conversations spent looking things up. the topics that would need a record of your own work stay small: knowledge retrieval and enterprise search at 3.61%, personal ai assistant at 2.86%, conversation and meeting intelligence at 0.26%.
here is the full topic mix, which makes the shape of the gap obvious.
| request topic | share of conversations |
|---|---|
| content creation and copywriting | 22.72% |
| education and learning | 13.23% |
| software development | 11.51% |
| research and intelligence | 10.94% |
| document processing and extraction | 4.32% |
| knowledge retrieval and enterprise search | 3.61% |
| personal ai assistant | 2.86% |
| conversation and meeting intelligence | 0.26% |
the demand is retrieval. the supply is nearly zero, and part of the reason is that people are not sure they are allowed to build the archive. mostly, on their own machine, for the pixels, they are.
it is worth noting who else has concluded the same thing. microsoft built recall into windows, openai and anthropic both shipped memory systems in 2026, and google rebranded gemini's context layer as personal intelligence in january 2026. every one of them is trying to solve retrieval. none of them solves it for what was merely on your screen.
how to do it defensibly
six things, and they take about 10 minutes to set up once.
- own the machine, or get it in writing. the cheapest protection available.
- default the audio off. turn it on per call, deliberately, after you have announced it.
- exclude aggressively. password managers, banking, health portals, hr systems, anything belonging to a client who has not agreed. a tool that cannot exclude by app and by site is not suitable for a work machine.
- keep the archive local. an archive that never leaves your disk is a much smaller problem under every regime discussed above, and it means no third party is holding other people's data on your behalf. it also survives an acquisition, which the rewind users of december 2025 did not.
- set a retention limit and honour it. "forever" is the hardest position to defend and the easiest to change.
- be able to delete on request. if a colleague asks not to be recorded, you need to be able to act on it.
remynd is built to sit inside those constraints. capture, ocr and storage stay on your mac by default, it captures the focused window rather than every display, and you can exclude specific apps or sites from capture entirely. sign in and the cloud agent do reach the network, which we state on the security page rather than claiming that nothing ever leaves the machine. the storage argument in full is in private ai on your mac.
the short version
recording your own screen on your own machine is, in most places, unremarkable. recording other people's voices without telling them is where the real exposure lives, and it is entirely avoidable by saying one sentence at the start of a call.
the questions worth answering before you install anything are: who owns this machine, whose data is on it, is the audio off by default, and can i exclude the things that must never be captured. get those four right and the legal question mostly stops being interesting. what remains is the harder one about where the archive lives, which we took apart in what happens to your data when an ai app shuts down.
again, and it bears repeating: this is orientation, not advice. laws differ by state and country and they move.
download remynd for mac and set your exclusions before your first recording.