Metraipolis › archive › dc_65e77153e80d

TRACE: Trajectory Aware Reasoning for Multi-Turn Adversarial Conversation Evaluation — arXiv:2608.15594

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.15594
License (as archived)
CC0-1.0
Archived
2026-10-18T17:25:56.182Z
SHA-256 of the archived text
883200f8f66fd3bc9fc84772950fb200b3764e1a8f112f024b3cad284a2aeeab
Stable URL of this page
https://metraipolis.com/documento/dc_65e77153e80d

How to cite

TRACE: Trajectory Aware Reasoning for Multi-Turn Adversarial Conversation Evaluation — arXiv:2608.15594 — arXiv (en), https://arxiv.org/abs/2608.15594 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_65e77153e80d (Metraipolis archive dc_65e77153e80d, sha256 883200f8f66fd3bc9fc84772950fb200b3764e1a8f112f024b3cad284a2aeeab)
BibTeX
@misc{dc65e77153e80d,
  title = {TRACE: Trajectory Aware Reasoning for Multi-Turn Adversarial Conversation Evaluation — arXiv:2608.15594},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.15594},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_65e77153e80d; sha256 883200f8f66fd3bc9fc84772950fb200b3764e1a8f112f024b3cad284a2aeeab},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15594. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multi-turn jailbreak attacks have emerged as a critical safety threat to LLMs, as harmful objectives are decomposed across a sequence of apparently benign turns to bypass guardrails. Existing defenses lack the reasoning capacity to identify evolving manipulation patterns, often trading helpfulness for safety by over-refusing benign requests related to sensitive topics. We introduce Trace, a multi-turn defense with trajectory-aware structured reasoning. Before generating each response, the model identifies manipulation cues from the trajectory, evaluates both the benign and adversarial interpretations of user intent, assigns a jailbreak score, and commits to an action: Allow, Caution, or Decline. We curate 4k multi-turn adversarial conversations from five attack frameworks, pair them with 2.4k benign dialogs, and 600 sensitive-but-benign conversations. We train Llama-3.1-8B-Instruct with SFT and GRPO under a multi-component reward that jointly optimizes helpfulness on benign prompts and robustness against jailbreak attempts. Across seven multi-turn attack benchmarks, Trace attains an average attack success rate (ASR) of 14.5% against 31.4% for the strongest baseline and 74.9% for the undefended target, while significantly raising the attacker effort required per successful jailbreak. Trace also balances usability and safety, achieving a 93.3% average compliance on over-refusal benchmarks.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.15594 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T16:29:13.815Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).