adversarial training
Adversarial training trains a model on deliberately corrupted inputs so it learns to resist tricks that would otherwise fool it. By exposing the system to these examples during training, the model becomes robust to small, intentional perturbations that might cause errors. This is like practicing against a tough opponent to prepare for unexpected moves.
The papers show that adversarial training is most effective when attacks are tailored to the specific way inputs are presented, when training enforces properties like cyclical monotonicity, and when the latent space used for training naturally separates relevant features. Across domains, methods that adapt adversaries, leverage suitable feature spaces, or optimize complete rollouts lead to stronger robustness and better generalization than standard adversarial training.
- Papers
- 9
- Released code
- 0
- First seen
- Oct 2026
- Latest
- Nov 2026
9 papers in the last two months, against 0 in the two before.
Start here
Three papers to read first, in the order they came out- 1Three-stage latent generator reaches 1.11 FID on ImageNet
A distill-then-refine latent rollout approach with a staged Transformer achieves FID 1.11 on ImageNet 256x256 without Fréchet losses.
Oct 2026 - 2LLM defenses against prompt injection are channel-conditioned and require deployment-aware training
Indirect prompt-injection defenses fail across presentation channels, while channel-aware adversarial training sharply reduces attacks without substantially sacrificing clean accuracy.
Oct 2026 - 3AI agents cut prompt injection success from 79% to 9%
SecPO combines a new training objective with strong simulated attacks to cut prompt-injection success while preserving performance on benign tasks.
Nov 2026
How fast it is moving
Often studied with
prompt injectionA security attack where malicious instructions are embedded into inputs to override an LLM's intended behavior.3 togetheragentic reinforcement learningTraining approach that lets LLM-based agents learn multi-step tool use and planning via reinforcement learning, often with dense or per-tool rewards.1 togetherflow matchingA generative training method that learns an ordinary differential equation mapping noise to data by matching velocity fields along probability paths.1 togetherWho is working on it
The papers
Most central to this idea first, not most recent.
- Big Techcs.CR
AI agents cut prompt injection success from 79% to 9%
SecPO combines a new training objective with strong simulated attacks to cut prompt-injection success while preserving performance on benign tasks.
UC Berkeley, Anthropic · Nov 2026
- Research Labcs.LG
Safety training can make chatbots refuse even after they start complying
Adversarial Self-Distillation lets a language model recover from accidentally starting a compliant response, improving jailbreak resistance.
University of Cologne, CISPA Helmholtz Center for Information Security · Oct 2026
- Academiccs.LG
LLM defenses against prompt injection are channel-conditioned and require deployment-aware training
Indirect prompt-injection defenses fail across presentation channels, while channel-aware adversarial training sharply reduces attacks without substantially sacrificing clean accuracy.
Mohamed bin Zayed University of Artificial Intelligence · Oct 2026
- Big Techcs.CL
Reinforcement learning yields user simulators that stay faithful and sound natural
Reinforcement learning with a jointly trained discriminator produces user simulators that are both task-faithful and natural.
Cornell University, Microsoft Research · Oct 2026
- Big Techcs.CL
Adaptive simulated training makes web agents more resistant to prompt injection
AdvSim2Real co-evolves tasks, attacks, and a web agent in a simulated world, raising task completion under unseen attacks by 33.6%.
Mohamed bin Zayed University of Artificial Intelligence, Amazon · Oct 2026
- Big Tech
Cyclical monotonicity enforces efficient adversarial transport in robust learning
Adversarial training violates optimal transport geometry; enforcing cyclical monotonicity via MPA or ICNN improves robustness.
EPFL, Apple · Oct 2026
- Chinese Techcs.CV
Three-stage latent generator reaches 1.11 FID on ImageNet
A distill-then-refine latent rollout approach with a staged Transformer achieves FID 1.11 on ImageNet 256x256 without Fréchet losses.
ByteDance · Oct 2026
- Chinese Techcs.CV
Fused DINOv3 features enable fast, faithful real-world image super-resolution
Super-resolution achieves best fidelity-perception trade-off when performed in the frozen fused latent space of DINOv3-L.
University of California, Berkeley, Alibaba Group · Oct 2026
- Chinese Techcs.CV
Small feature adapters remove visual domains while preserving scene content
A small adapter on frozen DINO features moves a 13-14% residue to translate domains and outperforms pixel-fed generators with fewer parameters.
Huawei, Graz University of Technology · Oct 2026
Concepts are extracted from each paper and reused across the corpus, so this page grows on its own as the desk reads.