When Guidance Becomes a Guard: Early Interception of Unsafe Inputs in Text-to-Image Generation

arXiv:2610.26452v1 Announce Type: new Abstract: Text-to-image diffusion models can produce unsafe content, while output filtering intervenes only after synthesis. We investigate whether native generation signals enable early detection of unsafe and adversarial inputs. We find that classifier-free guidance exposes strong safety signals in conditional responses, accessible through compact statistics and linear readouts. Based on this finding, we propose Conditional Semantic Trajectory Interception (CSTI), which aggregates activation contrasts between conditional and unconditional responses at early checkpoints for calibrated risk detection. CSTI fits concept-specific detectors using canonical concept and safe examples, without attack samples, intermediate image decoding, or generator modification. Across five generators, four adversarial attack collections, and two general NSFW datasets, CSTI achieves the strongest overall performance among six baselines at a matched 5% FPR. Early interception reduces generation latency by over 60% on intercepted paths. These results demonstrate the effectiveness of native conditional responses for early safety interception.

Related

• Decoupled Safety Control: A Safety-Control Algorithm for Training-Free Safety Guidance
• Harmful Content Generation in Text-to-Image Models: Capabilities and Moderation Limitations
• SaTPO: Text-to-Image Safety Alignment via Transition Preference Optimization
• SPOT: Selective Prompt Projection via Total Variation for Inference-Only Safe Text-to-Image Generation

Source: arXiv cs.CV | 2026-10-20