Search papers, labs, and topics across Lattice.
The paper introduces Constructive Safety Alignment (CSA), a paradigm shift from refusal-based safety mechanisms in LLMs to a more human-centric approach that actively guides vulnerable users towards safe and helpful outcomes. CSA incorporates game-theoretic anticipation of user reactions, fine-grained risk boundary discovery, and interpretable reasoning control. The implementation, Oyster-I (Oy1), demonstrates state-of-the-art safety among open models while maintaining high general capabilities, exhibiting strong constructive engagement and robustness against jailbreaks.
LLMs can move beyond simple refusals to actively guide vulnerable users towards safe outcomes, achieving state-of-the-art safety and robustness against jailbreaks.
Large language models (LLMs) typically deploy safety mechanisms to prevent harmful content generation. Most current approaches focus narrowly on risks posed by malicious actors, often framing risks as adversarial events and relying on defensive refusals. However, in real-world settings, risks also come from non-malicious users seeking help while under psychological distress (e.g., self-harm intentions). In such cases, the model's response can strongly influence the user's next actions. Simple refusals may lead them to repeat, escalate, or move to unsafe platforms, creating worse outcomes. We introduce Constructive Safety Alignment (CSA), a human-centric paradigm that protects against malicious misuse while actively guiding vulnerable users toward safe and helpful results. Implemented in Oyster-I (Oy1), CSA combines game-theoretic anticipation of user reactions, fine-grained risk boundary discovery, and interpretable reasoning control, turning safety into a trust-building process. Oy1 achieves state-of-the-art safety among open models while retaining high general capabilities. On our Constructive Benchmark, it shows strong constructive engagement, close to GPT-5, and unmatched robustness on the Strata-Sword jailbreak dataset, nearing GPT-o1 levels. By shifting from refusal-first to guidance-first safety, CSA redefines the model-user relationship, aiming for systems that are not just safe, but meaningfully helpful. We release Oy1, code, and the benchmark to support responsible, user-centered AI.
Tsinghua AI