X postSamuel CardilloOpinion
Uncensored model weights provide security researchers with raw capabilities for stress-testing AI safety boundaries, while fueling ongoing industry tension regarding closed, heavily guarded platforms.
The brief commentary underscores the friction between tightly aligned commercial LLM developers like Anthropic and community releases of unrestricted model weights. While framed as tooling for red teaming and vulnerability assessment, such releases continually challenge safety-first commercial strategies.
Samuel Cardillo
@CardilloSamuel · 15 August 2026
anthropic will be furious. t.co/2POG8UjU3t
2k likes12 replies604 bookmarks227.9k views
View on XSummary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.
Similar in content, found by the AI search.
Link:OrcaRouter
OrcaRouter has released an abliterated, block-FP8 quantized version of the multimodal model Qwen3.8-27B on Hugging Face. By orthogonalising the internal refusal direction, standard safety guardrails were removed specifically for red-teaming, interpretability, and AI safety research.
Link:Empero AI
Empero AI has released Qwythos-27B-v1, an open-weight reasoning model based on Qwen3.5-27B. The checkpoint preserves multi-token prediction (MTP), vision capabilities, and a 1M-token context window while specialising in terminal and agentic workflows.
Link:PocketAiHub
PocketAiHub has released an unofficial MLX derivative family of Qwen3.8-27B with suppressed refusal behaviour ('abliteration'). The consolidated repository provides quantized builds ranging from experimental 2-bit AWQ to full BF16 optimised for Apple Silicon.
X post:Mia
A Hugging Face repository provides GGUF and safetensors quantisations of the Qwen-Image 2.1 base model for local execution in ComfyUI, featuring no integrated safety checkers or content filters.
2.5kbookmarks138kviews
X post:Cantina
Cantina has introduced apex-flash-1, an open-weights model designed for security research. It was post-trained on real vulnerabilities identified by the team, alongside an abliterated version designed to reduce refusals in authorized workflows.
1.5kbookmarks270.4kviews
Link:Qwen
The Qwen team has released Qwen3.8-27B, a compact, dense vision-language model under the Apache-2.0 licence. The 27-billion-parameter model pairs Gated DeltaNet with conventional gated attention, handles images and videos natively, and introduces flexible reasoning controls.
Parts of this content were created with the assistance of AI.
