VS Code 1.117.0 Automatically Adds Copilot as Co-Author on Commit — Here's What Triggers It

VS Code v1.117.0 is adding Co-authored-by: Copilot <[email protected]> to git commits when Copilot inline suggestions (autocomplete) are used. A HN user discovered this when a PR they authored without actively using Copilot still had the co-author line. Investigations confirmed the trigger: accepting even a single character from Copilot's inline suggestion — like fixing a typo in a changelog — appends the co-author attribution.
How it Happens
- The feature is enabled by default with Copilot autocomplete (inline suggestions).
- If you accept any suggestion (even a single comma), the commit gets
Co-authored-by: Copilot <[email protected]>appended. - Manually writing commit messages or using Copilot's commit message generation does not trigger it — only inline acceptance does.
- The commit line reads: "Committed by USERNAME and Copilot".
Community Reaction
- Misattribution: Users argue that claiming authorship by default for minor autocomplete is deceptive, especially when not explicitly using Copilot for that commit.
- No Disclosure: The feature was not clearly communicated in the UI — only buried in release notes from the day prior.
- Opt-Out Available: The entire Copilot feature set can be disabled via
chat.disableAIFeaturesin settings, though no granular toggle for co-author exists.
What You Can Do
- Disable Copilot autocomplete entirely to prevent unwanted co-author tags.
- Set
chat.disableAIFeatures: trueto kill all Copilot features. - Review commit messages before push — the co-author line will appear in the commit body.
For developers relying on Copilot autocomplete, this means every accepted suggestion automatically watermarks your commit history. Consider whether you want that on your PRs before updating.
📖 Read the full source: HN AI Agents
👀 See Also

Atlassian Enables Default Data Collection for AI Training
Atlassian has enabled default data collection across its products to train AI models, according to a source posted on Hacker News with 312 points and 75 comments.

Claude-Code v2.1.91 adds MCP result persistence, shell execution controls, and multi-line deep links
Claude-Code v2.1.91 introduces MCP tool result persistence override via _meta["anthropic/maxResultSizeChars"] annotation supporting up to 500K characters, adds disableSkillShellExecution setting, and enables multi-line prompts in claude-cli://open?q= deep links with encoded newlines.

Anthropic Secures 300MW Compute at Colossus 1 with 220,000 NVIDIA GPUs via SpaceX Partnership
Anthropic announced a partnership with SpaceX to use all compute capacity at the Colossus 1 data center, gaining over 300MW and more than 220,000 NVIDIA GPUs within a month.

Apple Silicon Benchmark: Qwen3-VL Performance on M3, M4, and M5 Max for Vision LLM Classification
Benchmark results show Qwen3-VL vision LLM classification performance on Apple Silicon: M3 Max and M4 Studio are nearly identical for 8B models, while M5 Max is 75-83% faster. Memory bandwidth matters more for token generation than prefill in vision tasks.