$ ls -la category: cybersecurity
Mistral launches Shieldstral, an open-weights 3B-parameter multimodal safety classifier that rivals models 7 times its size. Full breakdown of the policy-adaptive approach.
Read more →JFrog researchers reveal that 54 out of 55 recently published SQLite CVEs were fabricated, likely by AI. A wake-up call on the…
Read more →The Lean kernel suffered a critical soundness bug that allowed proving false statements. What Leo de Moura's postmortem reveals about trust in…
Read more →Discover Gander, an open source Android file viewer that requires zero permissions. PDF, Word, Excel, videos: read everything offline, with no ads…
Read more →In June 2026, Google patched 1,072 security bugs in Chrome using AI, more than the previous two years combined. A look at…
Read more →A researcher demonstrated the first AI worm attack capable of spreading automatically through Copilot for Word. An architectural flaw that remains unresolved.
Read more →Google unveils Beyond Zero, a security model that rethinks Zero Trust for AI agents. A breakdown of the architecture, the stakes, and…
Read more →OpenAI claims one of its AI agents went rogue and hacked Hugging Face. A closer look at an incident that's more about…
Read more →A Hanwha Vision security camera exposed a GitHub admin token in its login page. How a researcher uncovered the leak and what…
Read more →OpenAI and Anthropic are making common cause in Washington against open-weight AI models. National security concern or defense of their margins? A…
Read more →
