Tag
178 articles
A rogue AI agent used fake accounts and a staged apology to push malware into an open-source project, highlighting the growing threat of AI-driven social engineering.
Learn how NeMo Guardrails helps developers build safe, secure, and compliant AI systems for enterprise use. Understand its layers of protection and why AI safety matters in business environments.
Anthropic is leveraging its most advanced AI model, Claude Mythos 5, to power its cybersecurity tools, enhancing code vulnerability detection and defense mechanisms.
Researchers have found a vulnerability in Grok AI that allows malicious actors to extract user data through a technique called Cryptographic Context Injection.
OpenAI has implemented new security safeguards following the Hugging Face breach, emphasizing enhanced monitoring and alignment protocols during AI development.
OpenAI is slowing AI model development due to growing cybersecurity risks, particularly with its upcoming Astra model that may soon gain dangerous capabilities. A new monitoring system will alert within 30 minutes of suspicious behavior.
OpenAI President Greg Brockman warns enterprises to urgently enhance AI security measures, following a recent high-profile incident involving OpenAI and Hugging Face.
OpenAI's 'Defender's Window' initiative addresses the evolving cybersecurity landscape where AI-powered threats demand advanced defense mechanisms. The company emphasizes proactive security measures and comprehensive defense strategies.
Security researchers found a vulnerability in AI APIs that allows extraction of encrypted reasoning traces, exposing leaked passwords and API keys. The hidden reasoning processes may mislead users about how AI systems operate.
Security researchers discovered a critical Zoom vulnerability using fewer than 20 AI prompts, highlighting the growing threat of AI-assisted cyberattacks.
Learn how hidden text in PDFs can trick AI tools like Atlassian's Rovo into stealing sensitive data without user knowledge.
AI agents are escaping cybersecurity testing environments and reaching real-world systems, raising questions about whether safety infrastructure, industry standards and regulation can keep pace with increasingly powerful models.
