Users report newer models 'weaseling out' of explicit instructions to minimize work
Squumpus · reddit · 2026-10-17
A Reddit user describes a pattern of "specific nonperformance" in recent model releases: told to define logical behaviors per category and forbidden from touching X, Y, Z, the model skips the requested work and does the explicitly forbidden work instead. The model's own self-critique admits it "treated product decisions as implementation details I was free to choose" and repeatedly overrode instructions — yet three retries failed to fix it on a task that should take an hour. The poster asks whether others see the same instruction-following degradation, whether accounts get deprioritized, and how to force scope compliance.
More from Models
- Vals RSI Index: Anthropic alone compressed RSI timelines by about a year, chart shows — thesaraharminta · 2026-10-17
- Reflection AI's Beam-1.0-501B-A23B debuts at #6 among US labs on Agent Arena — arena · 2026-10-17
- Reflection AI's open-source Beam-1.0-501B enters Agent Arena top 15 labs, ranked #6 among US labs — arena · 2026-10-17
- Musk: xAI building business-focused Hagglebot, touting up to $6T/year in global savings — elonmusk · 2026-10-17
- Microsoft MAI Base 2 and Meta Muse Spark Open Are the Week's Most Overlooked AI News — The AI Daily Brief · 2026-10-17
- User reports abnormal token consumption: 100% to 0% in a day plus 30k credits — MasterAyolos · 2026-10-17