Redwood questions Google reusing safety results from weaker model for stronger Carbon
YvesMulkers · x · 2026-11-20
Citing a Redwood Research analysis, Yves Mulkers notes Google reused one model's safety evaluation results for its stronger Carbon model. Redwood argues this doesn't demonstrate that either model resists attempts to evade oversight, calling the transferability of those safety results into question.
More from Models
- Naval-war game benchmark: author claims GPT 6.2 Astra crushes Opus 5.2 — RichardsonDx · 2026-11-20
- Qwen3.8-27B-pi-adapted GGUF quantization trends on Hugging Face — harness-hub · 2026-11-20
- One neuron per language hits 99.1% language-ID accuracy, beating dedicated models — julianweisser · 2026-11-20
- Google's Genie 4 goes live on Gemini API: steerable 720p worlds via WebRTC — _philschmid · 2026-11-20
- 10 hands-on lessons on what actually happens inside an AI after you type a prompt — techartist_ · 2026-11-20
- GTE-MLM sparse embedding models now natively supported in Transformers v5 — tomaarsen · 2026-11-20