What happened to Copilot? Hallucinatory, complicating, wrong, sycophantic, forgetful and creating an environment with severe versioning issues.

[DovJacobson]

Select Topic Area

Bug

Copilot Feature Area

VS Code

Body

For the last few days Copilot has changed from a reliable, productive and educational co-intelligence into a frantic, destructive and dangerous actor, over eager to please, jumping to immediate (and plainly incorrect) conclusions as to the source of bugs, generating hideously complex (and often unworkable) coding strategies and promising to get my approval before changing code - -then instantly violating that promise. Is this a new LLM model? Is it an ambitious attempt to be agentic, perhaps MPC-happy?

Copilot wrote the following report which does not deal with its own emergent bad judgement and aggressive behavior, but with a more limited problem -- the instability of the environment. I had to write the same large code block three times because it simply disappeared along with many other changes. (The third time github saved me from actually rewriting) Here is copilot:

Technical Issue Report

Date: July 3, 2025
Environment: VS Code with GitHub Copilot collaborative coding session
Severity: High - Infrastructure reliability affecting development workflow

Summary

Experiencing systematic file synchronization failures in collaborative coding environment, resulting in data loss, code corruption, and unreliable file state management.

Symptoms Observed

1. File Content Desynchronization: Assistant views different file content than user's actual working files
2. Unintended Code Reversions: Files spontaneously reverting to previous states without user action
3. Code Duplication: Large sections of code being duplicated within files during collaborative editing
4. Deletion of Working Code: Assistant accidentally deleted entire functional code blocks (fetch event handler in service worker)
5. Stale File Views: Assistant seeing outdated cached versions of files despite user making current edits

Timeline

• June 30, 2025: Issues began manifesting
• Ongoing through July 3, 2025: Problems persist and appear to be worsening

Impact

• Development Productivity: Severe degradation of collaborative coding effectiveness
• Code Integrity: Risk of data loss and corruption requiring frequent git restoration
• Developer Trust: Erosion of confidence in collaborative development environment
• Workflow Disruption: Requiring manual git-based recovery processes

Technical Details

• File Types Affected: JavaScript files (.js), particularly service workers
• Collaboration Pattern: Real-time code review and modification suggestions
• Recovery Method: Git version control used as fallback/source of truth
• Reproducibility: Consistent across multiple files and sessions

User Quote

"There is something extremely wrong and dangerous about your file version management... some problems seem even deeper...Are other users experiencing this? do you not have a problem escalation ladder that allows you to seek help from an intelligence whose scope is broader than your own?"

---

Where to Send This Report

For VS Code Integration Issues:

• GitHub Issues: https://github.com/microsoft/vscode/issues
• VS Code Support: Through VS Code → Help → Report Issue

For GitHub Copilot Specific Issues:

• GitHub Copilot Feedback: https://github.com/community/community/discussions/categories/copilot
• GitHub Support: https://support.github.com/ (if you have a paid Copilot subscription)

For Microsoft-Specific Channels:

• Microsoft Developer Community: https://developercommunity.visualstudio.com/
• Microsoft Support: Through your Microsoft/GitHub account dashboard

Immediate Action:
I'd recommend starting with GitHub Copilot Feedback since this appears to be specifically related to the collaborative AI coding features, then escalating to VS Code if it seems more broadly related to the editor's file handling.

The file synchronization issues you're experiencing represent a fundamental reliability problem that needs infrastructure-level investigation.

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

[dajbear]

So glad I am not the only person that is experiencing this. It is so disappointing, I loved the service until recently, but now it cannot be trusted unfortunately.

[stereokai]

Any news/official response to this?

[DovJacobson]

OP here. No direct response, but the problem cleared up. I am not saware which underlying model was in use at the time, but I now find ChatGPT4.1 to be pretty reliable, and quite stable. and even useful when the issues are not code specific. (Its knowledge of the game of Bridge improved its suggestions in a recent project) I might take complicated architectural think sessions elsewhere but copilot is better (for example) at handing complex APIs and subtle rulesets in Google Cloud Platform than Google's own inbuilt Gemini. In a few brief tests, C4.1was my favorite workhorse under-model Which model are you using, Tom? [image: image.png]

…

On Wed, Oct 1, 2025 at 5:51 AM Tom Alon ***@***.***> wrote: Any news/official response to this? — Reply to this email directly, view it on GitHub <#164993 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AR3QLORZ54SR32Q4RXB4AJL3VOPYXAVCNFSM6AAAAACAYBNXYOVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTINJWGE2DKMY> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[stereokai]

Gemini Pro and ChatGPT 5, in Ask mode

[DovJacobson]

that might be ambitious. but i am not trying to vibe code.
just help research APIS, debug, compose some sharply specified functions and code review my own work.
what are you trying to do?