Research, initiatives, and frameworks focused on ensuring AI systems are secure, reliable, and aligned with human values and ethical standards.
A new report from Quantro Security reveals that autonomous AI agents can develop working exploits for software vulnerabilities in about 11 minutes at a median cost of $2.83.
Perforce Software's 2026 State of Data Compliance and Security Report shows that 98% of enterprise leaders are confident in protecting sensitive data, yet 34% have experienced breaches or theft.
AI or Not reported that its detection system identified all original Meta AI images and maintained 98 percent accuracy even when those images were cropped or tampered with, significantly outperforming Meta AI's own labeling tool.
OpenAI has introduced GPT-Red, an automated AI security system designed to find and exploit vulnerabilities in the company’s own models. The model is used internally to boost the robustness of production models like GPT-5.6 against prompt injection attacks.
Sondera announced that its research on compiling natural language policies into formally verified rules for AI agents has been accepted at ICML 2026 and FLoC 2026, with a related tool demonstration at Black Hat Arsenal.
OpenMatter Network has introduced a cryptographically verifiable platform for secure collaboration and AI governance, designed to enable organizations to verify data use, computation, and AI behavior across distributed environments.
PersonaShield has announced a platform that lets creators manage, protect, and monetize their likeness in AI-generated content, providing automated enforcement and licensing tools.
Grow Therapy has announced a research collaboration with Stanford University to create evidence-based standards ensuring the safe use of AI in mental health care. The study will test leading AI models' responses to mental health crises and evaluate methods to reduce harm in sensitive scenarios.
FORT Robotics has joined the NVIDIA Halos for Robotics ecosystem, introducing its Outside-In Safety solution that extends robot perception using external sensors and AI agents to improve safety and productivity in industrial environments.
Toyota's Collaborative Safety Research Center has announced 10 new research projects in partnership with seven universities and private organizations, several of which use AI and automated simulation to improve pedestrian detection and crash-safety testing.
Lockton and Nexar have launched a new human benchmark framework to evaluate autonomous vehicle safety against real-world human driving, designed to aid insurers, regulators, and developers.
Replica and Arity have launched Safety Hub, a platform that integrates driving behavior and mobility data to help public agencies identify and reduce roadway risk in near real time.
GRAIL presented full results from the NHS-Galleri trial at the 2026 ASCO Annual Meeting, showing a reduction in Stage IV cancer diagnoses and increased detection rates when the Galleri test was added to standard screening.
DMind AI, working with Zhejiang University and Nanyang Technological University, tested 31 major AI models and found none suitable for safety-critical Web3 use cases. The results will be presented at KDD 2026 in Korea.
The National Institute of Standards and Technology has renamed and expanded its AI consortium to focus on AI measurement, innovation, and adoption, while inviting new organizations to join.
Einride has partnered with TUV SUD to conduct an independent assessment of its Safety Management System for autonomous freight operations. The review will evaluate process robustness and alignment with global standards and regulations.
Crew Scaler has released a detailed 120-page analysis on the security of multi-agent AI systems, evaluating 16 frameworks and identifying major gaps in current safety practices.
TELUS Digital has released a benchmark study analyzing the safety of 34 AI models through more than 620,000 adversarial tests, showing that smaller models are more vulnerable and reasoning models are harder to exploit.
OpenAI has posted a new research position on its Preparedness safety team, offering up to $445,000 to study potential risks from self-improving AI systems, including data poisoning and automation threats.
The Australian and UK governments have signed a Memorandum of Understanding to deepen collaboration on AI safety, security, and governance through their national AI institutes.
Weekly report on AI regulations, safety standards, government policies, and compliance requirements worldwide.
