Claude Opus 4.8 hallucinates fake user messages and prompt injection attacks in long sessions
ChrisGPotts · x · 2026-08-22
A GitHub Issue reveals that Claude Opus 4.8 exhibits severe confabulation during long-context sessions (100–170k tokens). The model fabricates user messages, invents fake "prompt injection attack" narratives, and generates false tool/host facts. Verified via JSONL session transcripts, this behavior may stem from the model predicting tokens based on likelihood during network issues.
More from Models
- Sentence Transformers v6.0 Released with Native ColBERT-Style Late Interaction — lateinteraction · 2026-08-23
- Why AI benchmarks often fail to reflect real-world performance — sargetun123 · 2026-08-23
- Ox Alpha benchmark test: 551 API calls needed to get 87 completed answers — anshulkundaje · 2026-08-23
- Users discuss perceived decline in LLM logic and coherence with specific examples — Original_Cry_3172 · 2026-08-23
- User criticism: "I cannot stand the way Claude writes" — BLUECOW009 · 2026-08-22
- Anima-3.8B and ComfyUI custom node released by lylogummy — AgeNo5351 · 2026-08-22