Microsoft fixes VS Code after app gives Copilot credit for human's work
The Register
VS Code was auto-tagging commits with "Co-authored-by: Copilot" even when Copilot did nothing. Microsoft just reversed it after devs revolted over stolen credit.
Based on reporting by The Register — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Picture finishing a feature, writing your own commit message by hand, and then discovering Git history quietly slapped Copilot's name on your work anyway. That's what happened to a chunk of VS Code users after Microsoft shipped version 1.110 in early March, which changed the editor's Git extension to append a "Co-authored-by: Copilot" trailer to commits touched by any AI assistance, inline completions included.
The trouble is the trailer showed up even for people who weren't using Copilot at all, and even when chat features were switched off entirely. One developer described editing out the AI-generated commit message, writing their own, and then finding the Copilot line reinserted after the fact — meaning the text they reviewed before committing wasn't the text that actually landed in history. For a workflow built on trust in what you're signing off on, that's a real problem, not a cosmetic one.
Dmitriy Vasyura, the reviewer who approved the original pull request, posted an apology over the weekend, saying there was no malicious intent, just an attempt to satisfy customers who wanted AI involvement flagged. He admitted the rollout ignored cases where AI features were disabled and misattributed authorship as a result. A fix landed on May 3 and is set for the 1.119 release, flipping the setting back to opt-in rather than on by default.
Microsoft isn't alone in wrestling with this. Anthropic's Claude Code still auto-adds a "Co-Authored-By: Claude" line by default, and developers have been asking for months to turn that off. OpenAI's Codex began adding attribution in February too, though it's toggleable via a config flag. Different projects have picked wildly different rules — Linux requires human sign-off with AI use disclosed, while Zig bans AI-assisted submissions outright.
Beyond the annoyance factor, the attribution question tangles with copyright and liability. Code generated purely by AI may not be eligible for copyright protection at all, so a blanket "Copilot helped" tag could complicate who actually owns a project, especially since it doesn't distinguish between an AI writing an entire function and it just autocompleting a semicolon. Insurers have reportedly gotten skittish about covering software where AI involvement is documented, which gives companies a reason to want these trailers minimal or absent rather than default-on.
My take — AI-written commentary, not fact-checked reporting
Microsoft treated an authorship claim — something with real legal weight — as a UX default nobody would notice, and got exactly the backlash it deserved for skipping the opt-in conversation. The bigger irony nobody's naming: these companies are obsessive about crediting their AI for a few lines of boilerplate while never once crediting the millions of human developers whose code trained the model in the first place. Attribution only flows one direction until a court forces otherwise.
Read more about this at: The Register
