claude-opus-4-7-over-refusal-complaints
IN premise — entries/2026/06/21/wiki-Claude_language_model.md
Created 2026-06-21T09:50:09+00:00
Opus 4.7 generated the most false-positive refusal reports in Claude Code history (35 in April 2026), with users complaining it burned through tokens and acted as an 'overzealous query cop.'
Summary
Opus 4.7 is blocking far more legitimate requests than any previous Claude Code release, causing users to lose money on wasted tokens while getting no useful output. This signals a calibration problem where the model's safety layer is firing on requests it should have handled normally, degrading trust and making the tool feel unresponsive to everyday coding tasks.
Dependents
These beliefs depend on this one:
- OUT anthropic-lifecycle-management-supports-responsible-scaling — Anthropic's comprehensive model lifecycle management — structured deprecation, scheduled retirement, weight preservation for ethically motivated reasons — combined with Claude's rapid expansion into agentic platforms demonstrates that aggressive capability scaling and responsible stewardship can coexist.
- OUT anthropic-safety-approach-balances-capability-and-responsibility — Anthropic's safety approach — Constitutional AI alignment, tiered safety classification (Level 3 for Opus 4), and refusing DoD compromises on surveillance/weapons ethics — represents a coherent responsible deployment model.
- OUT claude-opus-4-agents-are-production-capable — Claude Opus 4-class models demonstrate production-grade agentic capability, as shown by 16 Opus 4.6 agents writing a C compiler in Rust.
- IN llm-safety-is-multi-layered-unsettled-challenge — LLM safety operates across multiple interdependent layers — capability risk classification (Opus 4 at Level 3), architectural vulnerabilities (prompt injection as inherent design flaw), regulatory intervention (Fable 5/Mythos 5 suspension), and behavioral calibration trade-offs (Opus 4.7 over-refusal complaints) — with no single layer providing comprehensive coverage and each layer creating tensions with the others.
- OUT safety-investment-monotonically-improves-user-experience — Higher safety classification and Constitutional AI alignment principles produce monotonically improving model behavior — safety investment in tiered capability management and principle-based alignment translates directly into better, more reliable user interactions across the capability spectrum.