Stay informed with free updates
Simply sign up to the Work & Careers myFT Digest — delivered directly to your inbox.
There was a time, about six months ago by my count, when you could join an online meeting at work without having to wonder if it was being recorded or not.
It was such a basic rule of office etiquette to ask permission to log a work call that, unless I was specifically told otherwise, I assumed no surveillance was under way.
Not any more.
Cheap and easy AI recording and transcription apps are spreading at such a pace that what was once thought odd or rude is fast becoming normal.
It was some time towards the end of last year when I first noticed I was being invited to join online calls with outside contacts that were being recorded without any verbal warning from the host. I only realised the meeting was being logged because it said so on the screen.
Once, having failed to notice any alert, I only realised my thoughts had been recorded after I was emailed a transcript of the discussion.
Why is this happening? Perhaps because the rate at which online meetings are being recorded is rising faster than you might imagine.
Data from Roam, a US company that makes virtual office software for remote and hybrid workforces, shows that in January, an average 22 per cent of its users’ online calls were recorded. That’s up from 12 per cent in January last year and 9 per cent 18 months ago.
Roam has just over 22,000 users worldwide and they are logging everything from scheduled meetings to impromptu discussions and even the odd two-minute chat with a colleague. Recurring meetings are commonly set to record automatically.
And this is probably just the start of it, says Stanford University economics professor Nick Bloom, an authority on hybrid working who has been advising Roam for several years.
“The data shows relentless increases of transcription because AI has made this free and accurate and more valuable,” he told me last week. “I see this only rising and likely spreading beyond video calls to phone calls and in-person meetings.”
Not that long ago, I would have thought such a prospect impossible, not least because it is illegal to record phone calls in many jurisdictions. Now I’m not so sure.
The thing about recording work conversations is it can be extremely useful.
Consider being able to flourish the digital evidence that Janet, not John, came up with the idea that raised third-quarter revenues 20 per cent during a one-hour meeting that started at 3pm on March 7.
Likewise, people who had to miss an important meeting because of illness or vacation can use ever-improving AI notes to catch up later. I realise the dubious standards of many work meetings make these gains questionable, but they nonetheless exist.
None of this excuses the way we have allowed technology to drive our behaviour to the point that what was simple good manners yesterday is out the door today. And I suspect that for some users, the benefits of recording will struggle to outweigh the risks.
I say this because I know of at least two instances where a person has stayed online after a meeting and gossiped with a colleague about the uselessness of one participant, only to watch in horror as an AI agent has fired off a transcript of these unguarded thoughts directly to said participant, and everyone else in the gathering.
Even in less exciting circumstances, recording will inevitably make a lot of people less spontaneous and direct.
That’s not just because everything they say can be checked by colleagues. It is also the knowledge of how easy it is to email a meeting transcript to anyone, even someone outside the organisation.
And what if AI starts doing more than transcripts? Could it record how often people participate, or keep their cameras on, or look at the screen with interest? Could it use facial expressions and mannerisms to do psychological profiling and suggest who is worthy of promotion or dismissal?
Either way, AI companies seem likely to be among the biggest beneficiaries of all this recording, and the raft of material it offers to help train models.
Ultimately, as with so much else in our increasingly AI-driven world, AI recording and transcription looks likely to spread further and faster, while its human users haplessly look on. I would like to think things will change as the ramifications sink in, but considering recent technological history, I would not want to bet on it.
pilita.clark@ft.com
