A few weeks ago I showed the problem with AI in one image.
Since then, OpenAI has released an amazing new image generator. If you're a ChatGPT power user, ask it to generate an image of you at work.
More than 1,200 people have now shared their images with me. Send yours! I'm writing about the patterns in next week's nerd processor. Now back to this week's episode...
Signal in the noise
By now you've no doubt seen last week's news that top-ranking US officials discussed war plans in a Signal group chat. For good measure, they accidentally added a reporter to the chat. Yikes.
This is all kinds of bad, but it started me thinking: What channels do the rest of us use when we want something to be truly private?
What needs to be private?
Most of us aren't planning military maneuvers, but we all discuss topics at work that we'd prefer not to broadcast publicly, e.g.
- A sensitive personnel issue
- An imminent organizational change or labor action
- The details of a tricky customer relationship
- New IP that isn't yet ready for broad feedback
Some topics are inherently private, but there is increasingly another reason that people prefer to keep their content locked down: they don't want their professional IP to get stolen to train someone else's AI.
Is anything really private, though?
How many people use Snapchat thinking that their sensitive messages will disappear? Plenty of others use Signal and WhatsApp for encrypted conversations. But none of it is truly private. As our public officials discovered, it's straightforward to screenshot these apps and share conversations in another channel.
Even if you do choose a nominally secure medium for your private content, user error can still get you in the end. One time earlier in my career, I was complaining about my boss to a friend of mine over text. Except that oops! I didn't send the text to my friend after all. I sent it to my boss. I wanted the ground to open up and swallow me whole.
Historically, if your private content leaked, you might end up embarrassed. These days, though, there are other consequences. Even when your personal content is legally protected IP, it seems like anyone can use it as they see fit. Meta is just the latest major tech company being sued for 1) stealing other people's works to train their AI and 2) trying to hide the evidence. Maybe they will have to pay damages eventually, but by then, the stolen content will already be part of the borg.
Where am I writing, when I'm writing in private?
In any given day, I use lots of channels to say stuff about work. Daily, I write things in email, several Slacks, WhatsApp, Signal, iMessage, ChatGPT, Notes, and social media. Several times a week, I also write in Kit where I manage the nerd processor newsletter, Google Docs, and a host of other apps, including one proprietary comms tool for highly sensitive stuff.
The stakes of exposure vary widely. I don't particularly care if my texts about my kid's volleyball game or my comments in a tech WhatsApp show up publicly. But do I want my AI satire song about a famous tech influencer to show up on the open web? It's not like a top-secret plan to bomb Yemen, but ideally no, I do not.
The IP stuff is harder. I feel a little queasy knowing that all my public writing is being aggressively stolen to train BigTech's AI models. But like a teenager who takes for granted that their personal data will be used to target ads for the rest of time, I have more or less accepted that this is The Way Things Are Now. I don't like it, and it isn't legal. But the genie is too far out of the bottle for me to believe this is going to change. In so many ways, Big Tech owns the law now; I expect the law (or at least the enforcement of the law) will change to accommodate what they want. Sorry, I know that's a little bleak.
With that said, I went back and looked at how I used all my channels over the last month. Most things I've written for work are not especially secret. But doing a loose classification, I found nearly 100 things I've written in the last 30 days that I wouldn't want blasted out in public. It's interesting to see what I've written in channels that are not private or secure:
The bottom line: Most of my private comments aren't especially media-worthy. If they leak, it won't create an international incident. So maybe it doesn't especially matter that I'm sending most of them in insecure channels? But the content matters to me.
This really got me thinking how slippery the slope is. These are the channels we all use all the time for day-to-day comms. It's a hard habit to break, even when the stakes are way higher.
Where are you working in private?
Kieran
If someone sent you this nerd processor and you liked it, here's a direct subscribe link with all the back issues. If you love AI + work nerdiness, please consider leaving me a tip to keep nerd processor broadly available.
My latest data stories | Tell your own Viral Data Stories | nerdprocessor.com