@misc{indiciae8af7fd117f80,
  title = {Breaking Agent Backbones: Evaluating the Security of Backbone LLMs in AI Agents},
  author = {Julia Bazinska and Max Mathys and Francesco Casucci and Mateo Rojas-Carulla and Xander Davies and Alexandra Souly and Niklas Pfister},
  year = {2026},
  url = {https://arxiv.org/abs/2510.22620},
  note = {Source identifier: 2510.22620}
}