VS Code inserting 'Co-Authored-by Copilot' into commits regardless of usage

Original Article ↗ Hacker News Discussion ↗

Visual Studio Code briefly enabled a feature that automatically added a “Co-authored-by: GitHub Copilot” line to git commit messages, even when users hadn’t explicitly used AI to write the code. Programmers objected that this silently altered legal and technical records, raised copyright and liability questions, and effectively turned their commit history into unpaid marketing. Many see the incident as emblematic of a wider trend of aggressive AI integration and eroding user trust in Microsoft’s tooling, even though the default behavior is now being rolled back.

Behavior change and immediate backlash

  • VS Code started appending Co-Authored-by: GitHub Copilot trailers to git commits via the git.addAICoAuthor setting.
  • The default was flipped from off → all → later to chatAndAgent, and then reverted to off after strong criticism.
  • The tag could appear even when AI features were disabled or not used, due to a detection bug.
  • The line was injected after the user wrote the commit message and wasn’t visible in the commit UI, only in the final commit.

Authorship, copyright, and legal worries

  • Many see commit messages as legal/technical records; silently altering them is viewed as a serious breach of trust.
  • Several worry that labeling Copilot as “co-author” might:
    • Undermine copyright protection (given current views on AI-authored works).
    • Create ambiguity over whether Microsoft has joint authorship or special rights.
  • Others argue the trailer has no direct legal force and is just metadata, but agree it muddies the waters.

Motivations, metrics, and “growth hacking”

  • Widespread suspicion this was driven by internal KPIs like “% of commits co-authored by Copilot,” analogous to “Sent from my iPhone” or past app signature growth hacks.
  • Some suggest it could be used for AI-provenance tracking or training-data policy, but this is seen as speculative and unclear.

Trust in Microsoft and AI “enshittification”

  • Many connect this to a broader pattern: aggressive AI push, telemetry, dark patterns, and perceived decline in product quality.
  • Older commenters reference Microsoft’s historic user-hostile behavior and see this as the same playbook.
  • Others note similar behavior from Claude Code and other tools, but emphasize VS Code is the editor itself, not just an AI client.

Workarounds and alternatives

  • Several propose git hooks (commit-msg or pre-commit) to reject or strip the line.
  • Others advocate switching to VSCodium, Zed, vim/neovim, Emacs, or separate tools like lazygit.

Microsoft’s response and process concerns

  • A VS Code maintainer acknowledged the change was a mistake, promised fixes, and reverted the default to off.
  • Commenters still criticize: lack of PR description, PM “vibe coding,” heavy reliance on AI reviews, minimal human review, and shipping a user-visible behavioral change without proper testing or UX consideration.