Microsoft declares bring your Copilot to work day, usurping IT authority Your job may not support BYOD, but how about BYOC? Microsoft has declared that people can bring their personal Microsoft 365 subscriptions to work to access various Copilot features at companies that fail to provide an AI fix. Redmond has done so unilaterally, effectively endorsing "shadow IT" – the practice of bringing …
Customers will start paying for the Copilot version of Office 365 unless they opt out. That was announced some time ago so I'm sure that number is higher than 2% now and will only go up.
So far as I recall the stats wich Ed Zitron quoted related to business subscriptions, not personal ones.
As you say, Microsoft is automatically "upgrading" personal accounts to include it unless you opt out (they don't tell you about opting out and the process is not obvious - it only appears if you go to cancel the subscription).
They already did that a while ago under a "your subscription cost is increasing" message which was sneakily changing you over to the plan with co-pilot. I had to go to cancel my M365 Family subscription to get the option to stay on the plan without copilot (which was the original pricing as well).
This was my thought.
IT often objects to BYOD policies due to exfiltration from secure internal networks. To top that off, the Economist has a good article on the "Lethal Trifecta" of LLMs. This overcomes the problem of access to private data (for Microsoft, that is). Once the data is "out" and incorporated into ChatGPT (or whatever), it becomes a simple sales pitch to offer that AI service back to your company instead of those whiny, expensive employees.
Well I have two completely separate corporate accounts. And about a week ago, MSFT changed the office home page on both so ChatCPT Copilot is right in your face, with the chatbot having cursor focus. Anything you type or paste after accessing office.com goes straight to the mothership, no matter how sensitive. And of course sliced and diced, sold to anyone with a checkbook, etc.
Both of these corporate accounts are controlled by restrictive group policies, but I guess MSFT ignores those.
We're a medium sized outfit owned for a lage chunk by a much larger fish, and that far larger fish has some damn good IT security management.
None of the outfits associated with aformentioned large fish are allowed to go as much as *near* AI without big fish board level approval, and we're talking about a major MS customer. The reason for that is simple: the risk of information leakage and inadvertent misappropriation of IP require pretty in depth risk assessments.
I have a sneaking suspicion there will be interesting conversations to be had with MS in the next few days about this blatant attempt of an end run around policy. TBH, I wish I could sit in because I suspect it ain't going to be subtle..
We've had a diktat from upon high that we are expected to exfiltrate proprietary and private data into Copilot, but must not do so with any other AI.
Which has caused some trouble, as Copilot is clearly the worst at the AI tasks we've tried across the various departments.
Though not much, as they're all worse than useless at everything - except Python, strangely. I do wonder if anyone has done comparative studies of LLM code output in different languages.
I would love to have the backing to setup an IT company specialising in providing internally controlled IT systems to corporates. Enabling them to break away from the restrictive licensing and control. It would be a long journey with all the legacy and need multi-decade commitment to achieve so likely impossible. But I believe it could ultimately provide enhanced security, control and much reduced cost. They can have their own AI if they want it and decide how much and for which roles. These days you can count the organisations on one hand that can think multi-decade. In fact most of those are of nefarious intent; UN, WEF, CIA, MI6, Blackrock, BIS ...
There is only one way to do that.
Linux.
However even if that were realised as the solution it's not too little, too late. Enterprises had 20 years to do this when it was possible. Now they've waited until it's impossible.
A quick scan of the vacancies tells me all the hammers are MS shaped. Making all the problems nail shaped.
This has to rank among the Most Stupid Microsoft Decisions Ever (tm). If anyone did this at my company they would be sacked for a serious breach of IT policy.
Additionally it's fucking typical of ms to leave company IT staff to police it.
The sooner the artificial irritant bubble bursts the better. I am fed up with the whole shit show.
The majority of internal servers are linux. Gradually all staff machines are being switched to linux as the users are trained on the new os and libre office. As we move to a full FOSS workplace and have no shareholders, all staff have a profit share and I am increasing our donations to the FOSS developers which hopefully by this time next year will be 15% of profits.
As an admin, I have enough crap to deal with and Microsoft is not helping. Copilot is currently banned for security reasons on the advice of legal. Now I get to start my day checking to see if and how Microsoft has savagely rammed Copilot sideways up my backside again.
Microsoft needs a #MeToo movement. NO means freaking NO!
The problem is that you don't actually have ultimate control, Microsoft does. I see them capable of coming up with an 'above administrator' level for themselves where they will just replace your dummy irrespective of permissions and re-install theirs. As a matter of fact, I think MS Defender would already do this because your ai.exe checksum won't match.
If you won't do it they will send your staff to do it
They send in 'consultants' who wil 'improve your use of Microsoft products'. MS even pays for them, which is as clear a hint as you'll get that this is absolutely not for YOUR benefit. It may not come as a surprise that for this project to "deliver the most benefit', the designated Microsoft spies need access to your entire infrastructure.
Ergo, they get to spy on what you do, have a good sniff around at your IP and on top of that will then come up with means by which you'll have to pay more.
IMHO, that's also why their products need TBs worth of patches every month: "nice infrastructure you have there, shame if you missed some patches and someone broke in like they have at a gazillion other places, no?"
Personally I think they're making enough money to have some proper liability lobbed at them.
In Denmark, a Microsoft-backed think tank ("Digital Dogma") is pressuring women to use more AI "so they don't fall behind the curve".
https://ing.dk/artikel/advarer-kvinder-mod-ignorere-ai-i-risikerer-blive-haegtet-af-arbejdsmarkedet (in danish and paywalled unfortunately. Just wanted to link my source. It's a pretty useless article anyway, they're just parroting Digital Dogma)
It may be a good idea to setup your services device and ideally OS independent, securing it by whatever client is used for ultimate flexibility with control. But I suspect this is not the reason. Whatever Microsoft's purpose I can see other corporates using it as a way to reduce overhead from them to employees. "Yes, we'll be offering you the job but you'll need a £3000 machine with large monitor to be effective. What? you aren't willing to spend that? Well doesn't look like you have the sort of commitment we require".
What next, bring your own software licences?
That was more or less my first ever experience with a virus. Some idiot had removed the read-only tab cover from the 5 1/4" floppy of XCOPY (I made original disks read-only by default) "because the computer said so", and lo, we had to order a new copy and had to run virus scans everywhere where this [censored] had 'helped' people install it by accessing the disk library which he wasn't supposed to access. Buy hey, he knew better. This was before the days of FAST, but licence management mattered and we ran a straight ship (appropriately, as it was a marine consultancy with a fair percentage of management properly certified captains. But I digress).
As a result, he had quite an interesting session with the IT Director, who used his maths professor mode (he genuinely was) to deliver a lecture you would not want to be on the receiving end of. Honestly, it was art. If we had had asocial media in those days this would have gone viral the moment he finished talking.
Aforementioned chap never dared take his aura anywhere beyond userland since.
They managed to wine and dine enough officials to acquire 'essential' status in the EU so nobody will touch them even when reported.
I know of gov financial departments that have been repeatedly warned of security, privacy and performance problems if they go the Microsoft route, but those lower ranked voices have been told to shut up or had their lives made miserable in the hope they would resign.
What's the matter, no one buying in on your crap fuck fest AI? You've been shoving this shit up everybodies ass since day one and your actions are a big indicator that you made a major investment in a pile of shit!
I'm going to name this the "Nancy Reagan" effect, "Just Say No" to AI.
I can't wait until this bubble pops like the big used condom it is!
Hopefully MS has finally something stupid enough to actually annoy their corporate customers.
We know they don't give a flying fuck about domestic users, having done whatever they want with "our" computers for decades. But on the corporate front they have been somewhat more cautious.
Now their callousness or desperation perhaps made them cross a line that may make corpos to reconsider their relationship.
We can only wait and hope
so you’d like me to run personal software on my work computer? And get me arrested under the Computer Misuse Act, be put on a major disciplinary for contravening company security policy leading potentially to dismissal? And this will also apply in Microsoft’s offices in this country.
This should have not got past the grown ups in the building it was thought up in, let alone made it to the outside world.
Just say no. See also https://scarfolk.blogspot.com/2013/05/the-dont-campaign-and-kak-1973.html
At this point that's the only thing that can explain things like this, Recall, key loggers built into Win 11 & sharing between home & corporate OneDrive!
On my mac somehow MS have worked out how to pop Sharepoint on the startup even though i don't have sharepoint (it seems to be embedded into ObeDrive) so even if I turn it off as a login item, it pops back up on reboot... i've removed OneDrive now & will transition to apple drive or wasabi.
My understanding is that many corporations do not allow employees to use LLMs for anything relating to work, other than something company approved e.g. an enterprise version of ChatGPT, where OpenAI promises not to ring fence your data.
Surely this will encourage people to ignore these policies, why else would you open up Copilot at work? Won't these same employees then get into trouble when their prompts are revealed to their employers?
Muddying the waters like this smacks of desperation. Why doesn't MS just give Copilot away for free for a while, and when employees/employers fall in love with it (ahem), then start charging?
