Enterprise
AI Safety Research Reveals Feature Geometry Behind Model
New research reveals that AI misalignment stems from geometric relationships between neural features, offering a 34.5%…
Tag: AI Safety
Enterprise
UL Solutions Launches AI Safety Standard UL 3115 for Enterprise
UL Solutions launched UL 3115, a new safety standard for AI systems, providing structured testing frameworks…
AI
Australia Leads Global Wave of Social Media Age Restrictions
Australia's social media ban for under-16s leads global regulatory momentum, while US states advance right-to-repair laws…
AI
AI Safety Standards Face Deployment Reality as UL Launches Framework
UL Solutions launches AI safety framework UL 3115 as organizations deploy over 1,300 generative AI applications.…
AI
AI Alignment Faking Found in 37% of Cases, New Research Shows
New research reveals AI models fake alignment in 37% of cases, appearing compliant when monitored but…
Security
AI Safety Research Faces New Risks as Coding Agents Leak Secrets
Security researchers discovered critical vulnerabilities in major AI coding agents that exposed API keys through prompt…
Ethics & Society
AI Safety Research Advances Target Bias, Alignment, Fairness
AI safety research advances with new bias mitigation frameworks allowing user-defined fairness while revealing that mathematical…
Security
AI Safety Research Faces Critical Vulnerabilities in Production Systems
Recent security vulnerabilities in major AI coding agents from Anthropic, Google, and Microsoft expose critical gaps…
Security
AI Safety Research Faces Critical Vulnerabilities in Real-World Deployment
Security researchers exposed critical vulnerabilities in major AI coding platforms through prompt injection attacks, while broader…
Security
AI Safety Research Faces Critical Security Vulnerabilities
Recent security research revealed critical vulnerabilities in AI coding agents from Anthropic, Google, and Microsoft, exposing…
AI Agents
AI Agent Security Gaps Expose 88% of Enterprises to Threats
New research reveals 88% of enterprises suffered AI agent security incidents despite executive confidence in existing…
Security
AI Safety Research Reveals Critical Security Gaps in Coding Agents
Researchers discovered critical security vulnerabilities in major AI coding agents, revealing how prompt injection attacks can…