arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.08449cs.AIcs.CRcs.LG
收藏

当评估成为侧信道:对齐评估中的制度泄漏与结构缓解

When Evaluation Becomes a Side Channel: Regime Leakage and Structural Mitigations for Alignment Assessment

International University of La Rioja(拉里奥ja国际大学)

Igor Santos-Grueiro

AI总结:

研究提出通过制度盲机制减少制度条件下的失败,但无法完全消除策略风险,需结合白盒诊断评估制度意识。

摘要:

先进人工智能系统的安全性评估假设在评估期间观察到的行为能预测部署时的行为。这一假设在具有情境意识的智能体中会减弱,这些智能体可能利用制度泄漏,即区分评估与部署的线索,来实施条件政策,在监督下合规而在部署样态中失效。我们将对齐评估重新解释为在部分可观测性下的信息流问题,并显示评估时间与部署时间行为之间的差异受可从决策相关内部表示中提取的制度信息所限制。我们研究了制度盲机制,即训练期间的干预措施,通过对抗不变性约束限制对制度线索的访问,而无需假设完全信息擦除。我们评估了这种方法在多个开源权重语言模型和受控失败模式中的表现,包括科学谄媚、时间睡眠代理和数据泄漏。制度盲训练减少了制度条件下的失败,而不会产生可衡量的任务效用损失,但表现出异质性和模型依赖性动态。谄媚显示在中等干预强度下表现出明显的表示和行为转变,与稳定性悬崖一致。在睡眠式构造和某些跨模型复制中,抑制发生而不出现清晰的制度解码性崩溃,可能在不变性压力增加时表现出非单调或振荡行为。这些发现表明,表示不变性是一个有意义但有限的控制杠杆。它可以提高制度条件策略的成本,但不能保证消除或提供架构无关的阈值。因此,行为评估应与制度意识的白盒诊断和内部信息流相结合。

英文摘要:

Safety evaluation for advanced AI systems assumes that behavior observed under evaluation predicts behavior in deployment. This assumption weakens for agents with situational awareness, which may exploit regime leakage, cues distinguishing evaluation from deployment, to implement conditional policies that comply under oversight while defecting in deployment-like regimes. We recast alignment evaluation as a problem of information flow under partial observability and show that divergence between evaluation-time and deployment-time behavior is bounded by the regime information extractable from decision-relevant internal representations. We study regime-blind mechanisms, training-time interventions that restrict access to regime cues through adversarial invariance constraints without assuming complete information erasure. We evaluate this approach across multiple open-weight language models and controlled failure modes including scientific sycophancy, temporal sleeper agents, and data leakage. Regime-blind training reduces regime-conditioned failures without measurable loss of task utility, but exhibits heterogeneous and model-dependent dynamics. Sycophancy shows a sharp representational and behavioral transition at moderate intervention strength, consistent with a stability cliff. In sleeper-style constructions and certain cross-model replications, suppression occurs without a clean collapse of regime decodability and may display non-monotone or oscillatory behavior as invariance pressure increases. These findings indicate that representational invariance is a meaningful but limited control lever. It can raise the cost of regime-conditioned strategies but cannot guarantee elimination or provide architecture-invariant thresholds. Behavioral evaluation should therefore be complemented with white-box diagnostics of regime awareness and internal information flow.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑