We’ve developed a method to assess whether a neural network classifier can reliably defend against adversarial attacks not seen during training. Our method yields a new metric, UAR (Unforeseen Attack Robustness), which evaluates the robustness of a single model against an unanticipated attack, and highlights the need to measure performance across a more diverse range of unforeseen attacks.
Recommended
Operation "Doppelganger": Russian influence activity targeting UkraineOpenai Everyday work with ChatGPT WorkOpenai Romance scams: AI-enabled romance scam workflowsOpenai Will Hurd joins OpenAI’s board of directorsOpenai Statement from the OpenAI Board of Directors on the Nonprofit Commission ReportOpenai Simplifying, stabilizing, and scaling continuous-time consistency modelsOpenai
Selecting the Right Database for Your MicroservicesThenewstack
Supporting Kubernetes with Faster, Easier Test EnvironmentsPulumi
Develop, Preview, TestCss Tricks
Data Ethics Researcher Cautions Against Algorithmic Reordering of SocietyThenewstack
Lessons Learned from a Year of Testing the Web PlatformCss Tricks
Protecting Against CDN Cache PoisoningBunny
7 best MCP servers for Claude Code in 2026Mintlify
Introducing Socket Firewall: Free, Proactive Protection for Your Software Supply ChainSocket
The Problem with Transatlantic Data TransfersThenewstack
Beyond API Security Testing: Runtime ProtectionThenewstack
Introducing the pulumi policy analyze Command for Existing StacksPulumi
Unit Testing Cloud Deployments with .NETPulumi
Real-world Cyber Attacks Targeting Data Science ToolsAquasec
A “new direction” in the struggle against rightward scrollingCss Tricks
