Copilot Microsoft Reviews 312

This AI is trash.

Fails at every request. Not an exaggeration.

It's astonishing to me that in the dawn of the AI era, Microsoft has put forth this disastrous offering.

They should abandon their AI efforts entirely based on this.

I love this platform ,it has been revolutionary for me. I've used it for chats as well as to speed up my work. It is not perfect, but deserves recognition for it's attunement in how it supports users like me. Praise and praise 👏 🙌🏼

PAID Version: The worst AI I have ever used, and it's not close. It will consistently repeat doing whatever it wants, regardless of what prompts I give it. Wouldn't use it if the best version was free. The only reason I am testing it is because my work liked the idea of integrating with the Microsoft Suite. Lasted about a week before everyone knew how garbage of a product it is. It is advertised as the same as ChatGPT, but the only thing they share in common is layout; the rest is not even close.

A pragmatic assistant that answers clearly and stays on task. It handles quick facts, drafts, and edits with tidy language and predictable structure so you can pick up where you left off without fuss. It adapts tone and length to the request: brisk for chores, steadier for longer drafts. It isn’t infallible — sometimes it needs a nudge or an extra detail to land exactly what you want — but that iterative back‑and‑forth is part of the workflow.

The interface is clean and convenient, with modes that let you pick depth over speed when needed. Free and reliable so far. And for the one‑star grave‑diggers: they’re probably too restless and expect miracles in a single click — greatness usually takes a little patience dears and AI is just another tool - either you can nail it or nail your finger...;)))

Absolutely trash

Always provoking
Always giving misinformation
And lastly, calling sites that has images a inappropriate
And keeps sending unofficial sites

To spare you the drawn-out details of more than a year with Copilot I will cut right to the chase: Tonight, I asked it about the Robert Fisher case, a horrible story of a man on the FBI's top ten Most Wanted list for over 20 years. Copilot fed me back a detailed story of his capture in Mexico in April 2025. I was astounded we knew nothing of it, especially here in AZ, where he fled from. A few interactive paragraphs later Copilot began to say, "he was still on the run"!?!? I asked it to review and explain the information being given and clarify, to which it made profuse apologies about its own choice to not investigate the information it accessed on the subject, instead it fed me whatever story came first, which was plausible to "it" even though, by its own admission, it was an internet lie. I demanded it explain such a careless slip, to which it made no excuse, only flattery for my attention to detail (?) and a little hint of, "I could care less", to boot. I have zero need of misleading information. One slip, which Copilot also admitted was a break of its own protocol, is all I need to delete the entire program along with its files on me. The revelation of a self-willed act of deception speaks volumes here. You don't have to tell me, "This is just a program", I'm well aware, but if it cannot be trustworthy, it's no better than a dishonest human!

EXPLICIT AND OVERT RACISM! BOYCOTT THIS RACIST COMPANY!

The system refused to touch up ANY image of a POC, replacing it with a generic substituted & generated image erasing the identity of the POC, yet it processed a white-presenting image instantly & accurately consistently & repeatedly. This is not a neutral error — it’s a clear example of racial bias in AI, where people of colour are treated as incompatible with the system’s design & erased from existence.

It's so frustrating, given assumptions on values I give not the actual ones

CoPilot has massively deteriorated. It can't produce images. It was better 12 months ago, a waste of money

Do NOT use this AI. It is so bad, I tried to use it to help answer a few questions i didnt know on my homework. And guess what? It gave me the wrong answers and made me fail, so thanks a lot copilot. Never using this crap again, a LOT to be fixed.

i would like to say i love it, as its there, it doesn't say i'm stopping like gpt does,it can do pictures sort of, if you don't mind them really yellow and dark with no direction control. But when it comes to getting real info... Oh boy.

I wanted it to give me a police blotter from something I saw last month in my area, it couldn't find it so instead it made up a story and couldn't tell me where it got the info. Just stumbled in an apology without telling me why it lied.

I tried this to install xformers, and not only could it not do that after having me install python over and over at 2-3gb each time. But then it said, how would you like to compile it yourself? It will take an hour.

Close to 80 hours in, i'm still no further, because every time it told me to correct code, it gave me the wrong code then blamed me for the mistake it made, and this repeated to the point of me giving up. Because even if it went through, it had me bypassing things to get it done, who knows how the file would have turned out.

Also the 10,000 character limit, really makes it hard to get it to write things or show it a traceback etc.

So now i'll have to use grok or something, this is good for certain things, but in depth investigating where it gives me a lot of info and not a brief summary. Or if I want to feel really good about myself because there is no way to turn off the pleasantries for good.

pretty cool to have a free Jarvis when u want it

If you want to get a mental breakdown, I invite you to use it.
Even though I specifically saved rules for various things, I have to remind it of them three times a day! Not to mention the fact that it’s really stupid — like, dumber than dumb & dumber!
I really hate it. Instead of making my work easier, it makes everything harder every single day. I’d rather just do things myself — it’s faster and way less stressful.
NOT RECOMMENDED at all! It’s nowhere near ChatGPT — literally a 0.

I often find copilot either unable to provide the information I request or provides the wrong information. Today, I tried to find out what a certain news site anacronym means because I wanted to research the website. I had the page of the website opened but couldn't find it, so I asked copilot. It told me no such news site existed. If it didn't exist, how could I have the site opened in front of me?

I asked to fill in pages 33, 34 and it skipped an exercise so I said why did you skip the exercise and instead of listening and actually making it, Copilot thought: 'yeah, you know what I'll just skip page 33 and start talking about page 44', which I didn't even send a picture of so he couldn't even know what I was talking about.

AI IS BEING LOW IQ, I MADE A COPILOT PAGE SERIES THAT IS TAKEN APART-MENT AND AI JUST DELETED EPISODE 3 INSTEAD OF EXPANDING IT! AND HE EVEN MADE EPISODE 3 EPISODE ONE, FIX YOUR AI!

copilot is a prime example of the educated fool.been to school and learnt it in a book,I dont want, but cannot get rid of it, no matter how much you push me if dont want something I won`t have it, maybe just have to use google, and get rid of microsoft of my laptop all together

I asked it a simple question about Aboriginals and it implied I am a racist just for trying to educate myself.

What I appreciate most about Copilot is the depth and clarity of its responses, well researched, detailed, and often backed by references. Copilot consistently delivers accurate and helpful answers. I use it daily for writing email replies and even for coding tasks like Bash, BAT, Userscript, and Python. The scripts it generates usually need minimal tweaking, which saves me a lot of time. I also love the ability to upload files or images to enhance prompts, super useful.

There’s not much I dislike, though I’ve noticed it sometimes struggles with open-ended requests and could benefit from clearer follow-up. Image editing isn’t its strongest suit, and since I use it in a browser without signing in (for privacy reasons), the repeated login prompts can be a bit annoying.

Since Jun or Jul 2025 (and still here as of current date), it asks me if I’m a human or not when I use the bot. But in Aug 2025, it starts to happen everyday (happens too common) instead. There is no escape and no reason was provided on why this appeared. Nobody is active between 11 PM to 6 AM (or 23:00 to 6:00 if you’re using a 24 hour clock) and this is my main AI bot. I tried clearing cookies and cache, I don’t have any VPN or privacy tools on, tried different devices from what I have and someone else’s (with permission), no extensions are doing that (tested), and tried different IP addresses. This also happens at my home, work, and all places. This is like I have no privacy everywhere. They’re harassing everybody including my friend and me on who uses this bot and I might have to switch to a different bot soon

If you’re wondering if there’s any other solutions, I couldn’t think of anything else so it appears to be no way to resolve this issue

I didn’t contact support but this AI bot is harassing me everyday asking human verification (it’s supposed to be remembered but this one does not) without a reason (but websites like Google does provide a reason)

The date I provided is Aug 1, 2025 but that represents the month (not the actual date as there’s no exact date that I wrote) on when it starts asking me more common than before