{
  "id": 14088,
  "url": "https://arxiv.org/abs/2607.25857",
  "title": "Shieldstral",
  "summary": "We introduce Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier that matches or outperforms models nearly 7times its size on text safety benchmarks and sets a new state of the art on multimodal safety classification. Shieldstral formulates content moderation as a binary question-answering task. This simple formulation unifies diverse moderation tasks into a single yes/no problem, enabling heterogeneous safety datasets with divergent taxonomies to be consolidated under one t",
  "authors": "Antonia Calvi, Avinash Sooriyarachchi, Giada Pistilli, Guillaume Lample, Maarten Buyl, Maximilian Augustin",
  "category": "research",
  "topics": "regulation",
  "orgs": null,
  "regions": null,
  "published_at": "2026-07-27T20:00:00.000Z",
  "fetched_at": "2026-07-29T05:10:12.205Z",
  "source_slug": "hf-daily",
  "source_name": "HuggingFace Daily Papers",
  "source_homepage": "https://huggingface.co/papers",
  "ethics_ai_record_url": "https://ethics.ai/record/14088",
  "original_url": "https://arxiv.org/abs/2607.25857",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}