Microsoft has rolled back a controversial change to its popular code editor after developers complained that it was incorrectly crediting artificial intelligence for work completed entirely by humans. The issue centres on Git commit messages in Visual Studio Code, widely used by developers across the UK and globally.
Developers Push Back on Default AI Attribution
In March, Microsoft introduced an update to VS Code (version 1.110) that automatically appended a “Co-authored-by: Copilot” line to commits involving AI-assisted code. The feature was tied to its AI coding assistant, GitHub Copilot, and intended to improve transparency around AI contributions.
However, developers quickly reported that the attribution appeared even when Copilot had not been used — and, in some cases, when AI features had been explicitly disabled.
One developer noted on a GitHub discussion thread that despite manually editing a commit message to remove AI-generated text, the final commit still included Copilot as a co-author. This raised concerns that the software was modifying commit metadata after user review — a significant issue in professional development environments where accuracy and traceability are essential.
Microsoft Acknowledges Misstep
Over the weekend, Dmitriy Vasyura, a reviewer involved in approving the original change, issued a public apology. He stated that the intention had been to support functionality expected by some users working with AI-generated code, rather than to misrepresent authorship.
Vasyura acknowledged that the implementation failed to respect user settings and should not have overridden disabled AI features. A fix was introduced on 3 May, with the correction set to be included in the upcoming 1.119 release of VS Code.
The revised approach restores the attribution feature to an opt-in setting, allowing developers to choose whether to include AI credit in their commits.
Wider Industry Debate on AI Attribution
The controversy highlights a broader debate within the software industry, including among UK-based developers and firms, about how to document AI involvement in code creation.
Other tools have faced similar scrutiny. Developers using Claude Code previously raised concerns about automatic attribution, while OpenAI Codex introduced a default attribution feature earlier this year, which can be disabled through configuration settings.
Open-source projects have taken differing approaches. The Linux kernel project requires developers to explicitly document AI assistance, while the Zig programming language project prohibits AI-generated contributions altogether.
Legal and Commercial Implications
The inclusion of AI attribution in code commits is not merely a technical issue — it carries potential legal and commercial consequences.
Under current intellectual property frameworks, including those recognised in the UK, purely AI-generated content may not qualify for copyright protection. This raises questions about ownership when AI tools contribute to software development.
There are also implications for liability. Some insurers have reportedly been hesitant to provide cover for businesses that rely heavily on AI-generated code. Explicit attribution could therefore complicate claims or give insurers grounds to deny coverage.
Additionally, a simple attribution line does not clarify the extent of AI involvement. It may be unclear whether an AI tool generated substantial portions of code or merely provided minor autocomplete suggestions — a distinction that could prove significant in legal disputes.
Balancing Transparency and Developer Control
The episode underscores the tension between transparency and developer autonomy. While some argue that AI contributions should always be disclosed, others maintain that such decisions should remain firmly in the hands of developers.
In the case of VS Code, the main frustration stemmed not from the existence of AI attribution, but from its automatic application without user consent.
As AI tools become increasingly embedded in software development — including within the UK’s growing tech sector — establishing clear, consistent standards for attribution is likely to remain a pressing issue.
Conclusion
Microsoft’s swift reversal reflects the importance of trust and control in developer tools. While AI-assisted coding continues to evolve, the industry is still grappling with how best to acknowledge its role without undermining human authorship or introducing legal uncertainty. For now, giving developers the choice appears to be the preferred path forward.
“Writer. Amateur musicaholic. Infuriatingly humble zombie junkie. General internet maven. Bacon enthusiast. Coffee nerd.”