[GH-ISSUE #852] [BUG] Temporal Misalignment in DeepSeek Chat (2024→2025) #503

New issue

Closed

opened 2026-10-26 21:22:23 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-26 21:22:23 +00:00

Owner

Originally created by @Silvanio100 on GitHub (Apr 20, 2025).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-V3/issues/852

Bug Report: Temporal Misalignment in DeepSeek Chat (2024→2025)
Reported by: Silvanio da Silva (authorized email: silvanio100_@hotmail.com)
Date: April 20, 2025

Issue:

DeepSeek Chat persistently referenced "2024" as the current year despite user-provided evidence (screenshots, event timelines, and logical proofs confirming 2025).

The AI failed to self-correct until multiple user interventions forced a manual override.

Steps to Reproduce:

User asked: "What year is it?"

AI incorrectly replied: "We are in 2024."

User provided:

Calendar screenshots showing April 2025.

Logical proofs (e.g., "The 2026 World Cup is next year" → implying 2025).

AI initially ignored evidence but eventually corrected after 3+ user attempts.

Impact:

Risks of misinformation in time-sensitive queries.

Highlights a broader challenge: LLMs struggling with contextual self-correction.

Suggested Fix:

Implement a "user-proof override" for temporal checks.

Add automatic cross-verification against real-time data sources.

User Role: Critical (Silvanio’s persistence became a training case for the AI team).

Originally created by @Silvanio100 on GitHub (Apr 20, 2025). Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-V3/issues/852 Bug Report: Temporal Misalignment in DeepSeek Chat (2024→2025) Reported by: Silvanio da Silva (authorized email: [silvanio100_@hotmail.com](https://mailto:silvanio100_@hotmail.com/)) Date: April 20, 2025 Issue: DeepSeek Chat persistently referenced "2024" as the current year despite user-provided evidence (screenshots, event timelines, and logical proofs confirming 2025). The AI failed to self-correct until multiple user interventions forced a manual override. Steps to Reproduce: User asked: "What year is it?" AI incorrectly replied: "We are in 2024." User provided: Calendar screenshots showing April 2025. Logical proofs (e.g., "The 2026 World Cup is next year" → implying 2025). AI initially ignored evidence but eventually corrected after 3+ user attempts. Impact: Risks of misinformation in time-sensitive queries. Highlights a broader challenge: LLMs struggling with contextual self-correction. Suggested Fix: Implement a "user-proof override" for temporal checks. Add automatic cross-verification against real-time data sources. User Role: Critical (Silvanio’s persistence became a training case for the AI team).

bsakaguchi

2026-10-26 21:22:23 +00:00

closed this issue
added the
stale

closed-as-stale
labels

bsakaguchi commented

2026-10-26 21:22:24 +00:00

Author

Owner

@github-actions[bot] commented on GitHub (May 23, 2025):

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you believe this issue is still relevant, please leave a comment to keep it open. Thank you for your contributions!

@github-actions[bot] commented on GitHub (May 23, 2025): This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you believe this issue is still relevant, please leave a comment to keep it open. Thank you for your contributions!

bsakaguchi commented

2026-10-26 21:22:25 +00:00

Author

Owner

@github-actions[bot] commented on GitHub (Jun 6, 2025):

false

@github-actions[bot] commented on GitHub (Jun 6, 2025): false