{
  "id": 4207,
  "url": "https://arxiv.org/abs/2605.17204v1",
  "title": "Event-Grounded Sparse Autoencoders for Vision-Language-Action Policies",
  "summary": "Vision-Language-Action (VLA) policies translate language and visual inputs into robot actions, where their hidden representations directly shape closed-loop behavior. However, mechanistic interpretability tools from language and vision-language models do not transfer cleanly to VLAs: outputs are robot actions rather than human-readable tokens, and interventions can only be tested via expensive closed-loop rollouts. We propose an event-grounded interpretability pipeline that anchors SAE feature a",
  "authors": "Xinchen Jin, Aditya Chatterjee, Pranav Kumar, Rohan Paleja",
  "category": "research",
  "topics": "safety-alignment,agents-autonomy",
  "orgs": null,
  "regions": null,
  "published_at": "2026-05-17T00:20:17.000Z",
  "fetched_at": "2026-07-14T16:30:50.572Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/4207",
  "original_url": "https://arxiv.org/abs/2605.17204v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}