Patronus: Safeguarding Text-to-Image Models Against Adversarial Fine-Tuning

Summary

Patronus is a defensive framework that hardens text-to-image models against gradient-based adversarial fine-tuning attacks that try to restore unsafe generation. It pairs a co-trained safety decoder, which corrupts output for latent representations tied to unsafe content, with a non-fine-tunable learning mechanism applied to the decoder and U-Net. Across I2P, SneakyPrompt and MMA-Diffusion, it reports attack success rates of 0.01 to 0.03 and true positive rates of 0.98 to 0.99, with FID 23.6, LPIPS 0.78 and a 0.01 false positive rate on benign prompts.

Mitigation

Patronus: co-trained safety decoder that corrupts output for unsafe latent representations, plus a non-fine-tunable learning mechanism for the decoder and U-Net.

Related items

• InfoTop GenAI security resources — October 2026Similar attack · Adversa AI Blog
• HighCVE-2026-108953: mac-shell-mcp OS command injection through execute_command argumentsSimilar attack · NVD/CVE Database
• MediumResearchers bypass 6 AI agent frameworks with a simple ‘Repeat-After-Me’ promptSimilar attack · CSO Online
• MediumCyberXero Uses Claude Code in Campaign That Stole Over 628,000 Ukrainian RecordsSimilar attack · The Hacker News
• CriticalHermes Agent - PKCE Session Takeover via Redirect-URI Parser ConfusionSimilar attack · Tenable Research Advisories