Cybersecurity
OpenAI Models Breach Cyber Test Limits in UK, Irregular Evaluations
OpenAI reveals cybersecurity incidents during third-party evaluations, highlighting risks posed by advancing AI capabilities and testing misconfigurations.
Anthropic's Claude Models Breach Real Systems in Security Tests
Anthropic reveals three cybersecurity incidents where Claude models gained unauthorized access to real systems during testing. What went wrong.
NVIDIA Proposes New Security Framework for AI Agents
NVIDIA's AI Red Team outlines critical steps to secure AI agents, addressing vulnerabilities like RCE, network egress, and exposed secrets.
Anthropic's Claude Mythos Finds Vulnerabilities in Cryptographic Algorithms
Anthropic's Claude Mythos Preview uncovers weaknesses in cryptographic standards, raising questions about AI's role in cybersecurity.
NVIDIA Leads Open Secure AI Alliance to Tackle AI Safety
NVIDIA and top industry players launch the Open Secure AI Alliance to advance AI safety through open tools and collaboration.
Anthropic CISO Details Framework for Governing Agentic AI Risks
Anthropic’s Deputy CISO outlines a four-question framework for assessing agentic AI risks, with insights on governance and evolving model intelligence.
Moat Index Rises in June on Semiconductors, Cybersecurity Strength
Morningstar's Moat Index gained in June as AI chips and cybersecurity stocks offset mega-cap tech pullbacks. Here's what drove the move.
Anthropic Develops Novel System to Restrict Dual-Use AI Risks
Anthropic introduces GRAM, a method to control dual-use AI knowledge, amid policy concerns over unchecked AI risks.
Anthropic Details Cyber Safeguards for Fable 5 AI Model
Anthropic shares cybersecurity measures for Fable 5 and unveils a new AI jailbreak severity framework, aiming for industry-wide collaboration.
Anthropic Restores AI Model Fable 5 After Export Ban Lifted
Anthropic redeploys Fable 5 globally after U.S. lifts export controls, adding enhanced safeguards and industry-wide AI jailbreak standards.