AI Wiki — Knowledge Base

Mistral releases Shieldstral, a 3B multimodal safety classifier that it says matches models up to 7x its size on text safety, available under Apache 2.0 (Mistral AI Blog)

Mistral AI Blog: Mistral releases Shieldstral, a 3B multimodal safety classifier that it says matches models up to 7x its size on text safety, available under Apache 2.0 — Every product that ships a m

model-releasestechmeme

4 Aug 2026

Search: “safety”

Full-Body Dynamic Safety for Robot Manipulators: 3D Poisson Safety Functions for CBF-Based Safety Filters

OpenAI's head of safety, Johannes Heidecke, is leaving as OpenAI integrates its research and safety teams; Mia Glaese will become VP of research and safety (Maxwell Zeff/Wired)

Deep QP Safety Filter: Model-free Learning for Reachability-based Safety Filter

Safety, or Just Capability? A Validity Audit of Agent-Safety Benchmarks

Few Tokens, Big Leverage: Preserving Safety Alignment by Constraining Safety Tokens during Fine-tuning

Cooptimizing Safety and Performance Using Safety Value-Constrained Model Predictive Control

Who Bridges Safety? Identifying and Targeting Cross-Lingual Shared Safety Pathways

CASA: Classification Augmented with Safety Attention for Robust Multimodal Alignment

Permissive Safety Through Trusted Inference: Verifiable Belief-Space Neural Safety Filters for Assured Interactive Robotics

Approximating Safety Feedback Without a Safety Oracle via Model Predictive Control

Reducing the Safety Tax in LLM Safety Alignment with On-Policy Self-Distillation

Distributionally Robust Safety Under Arbitrary Uncertainties: A Safety Filtering Approach

Why do no 'AI safety' people ever say how LLM AIs failed to live up to the hype? Why do the safety people never share the failures of LLMs? …

AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin

Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-Tuning and Can Be Mitigated by Machine Unlearning

A look at the UK's AI Safety Institute, whose researchers probe AI models for safety gaps, as its work becomes a blueprint for other governments' AI policies (New York Times)

OpenAI endorses the Kids Online Safety Act and Illinois SB 315, an AI safety bill to create requirements around transparency, incident reporting, and more (OpenAI Global Affairs)

Safety-aware Goal-oriented Semantic Sensing, Communication, and Control for Robotics

Boundary Sampling to Learn Predictive Safety Filters via Pontryagin's Maximum Principle

Initiation Safety: A Missing Dimension in Generalist-Robot Safety

Dynamic Optimization and Safety Indicator Injection for Jailbreaking Text-to-Image Models with Multimodal Safety Filters

Adjustment Speed as a Safety Constraint for Nonstationary Reinforcement Learning

When Context Flips, Safety Breaks: Diagnosing Brittle Safety in Aligned Language Models

Safety-Critical Whole-Body Control for Humanoid Robots via Input-to-State Safe Control Barrier Functions

Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time Alignment

Safe Control using Learned Safety Filters and Adaptive Conformal Inference

Dialogue based Interactive Explanations for Safety Decisions in Human Robot Collaboration

ProbGuard: Calibrated Safety Risk Estimation from LLM Output Distributions

Beyond 'I Can't Help With That': How Child Safety Experts Evaluate AI Chatbot Safety

SafeGen: Goal-Conditioned Video Diffusion of Safety-Critical Scenarios for VLM-Based Autonomous Driving

LIBERO-Safety: A Comprehensive Benchmark for Physical and Semantic Safety in Vision-Language-Action Models

Who Earns the Safety? Intervention-Aware Quantum Predictive Control with Safety Attribution

Do Models Share Safety Representations? Cross-Model Steering for Safe Visual Generation

Nemotron 3.5 Content Safety: Customizable Multimodal Safety for Global Enterprise AI

Safety Under Scaffolding: How Evaluation Conditions Shape Measured Safety

Safe Continual Reinforcement Learning under Nonstationarity via Adaptive Safety Constraints

Safety Is Not Universal: The Selective Safety Trap in LLM Alignment

Dual Stress: Runtime Safety Monitoring for Safety-Constrained MPC Navigation

SafeCap: Improving LVLM Safety with Image Captioning Reinforcement Learning

Mistral releases Shieldstral, a 3B multimodal safety classifier that it says matches models up to 7x its size on text safety, available under Apache 2.0 (Mistral AI Blog)

A profile of Jacob Tsimerman, who won the Fields Medal last week and is taking a leave from the University of Toronto to join OpenAI and work on AI safety (Ben Cohen/Wall Street Journal)

Configurable Reward Model for Balanced Safety Alignment

Why Do Safety Guardrails Degrade Across Languages?

Sustaining AI safety: Control-theoretic external impossibility, intrinsic necessity, and structural requirements

RACC: Representation-Aware Coverage Criteria for LLM Safety Testing

The Safety-Aware Denoiser for Text Diffusion Models

Dyna-Style Safety Augmented Reinforcement Learning: Staying Safe in the Face of Uncertainty

State-Dependent Safety Failures in Multi-Turn Language Model Interaction