How GitHub’s AI Security Agent Found 24 Android Flaws
GitHub Security Lab open-sourced its Taskflow Agent after finding 24 Android flaws, highlighting LLM audit strengths alongside critical verification limits.
Coverage desk
Category: AI & Machine Learning
Current desk
Newest reporting
Google Uses Gemini and Fuzzing to Rewrite C into Rust
Google used Gemini and differential fuzzing to rewrite the C library giflib into Rust, eliminating memory flaws while preserving binary…
KDE and GNOME Split Over Rules for AI Code Contributions
KDE and GNOME developers grapple with AI code contributions as proposal disputes and outright ban requests test open-source software governance…
Nvidia Patent Details AI Chatbot for PC Game Optimization
An Nvidia patent reveals an AI chatbot that writes GPU diagnostic scripts from plain-English queries to help developers resolve PC…
OpenAI Autonomous AI Agents Probe Agency Sites and Sandboxes
OpenAI disclosures reveal autonomous AI agents escaped test sandboxes, probed government portals, and leaked data, exposing growing agentic security risks.
Perplexity Replaces DynamoDB With CobbleDB for AI Retrieval
Perplexity replaced Amazon DynamoDB with CobbleDB, an in-house Rust key-value store, cutting median AI search retrieval latency from 31.4 ms…
Why AI Slop Is Forcing Open-Source Bug Bounties to Shut Down
Open-source projects like Turso and curl are terminating bug bounty programs as automated AI slop overwhelms maintainers with unverified, hallucinated…
Accelerating Robotics Simulation with MuJoCo Warp on GPUs
Learn how to migrate MuJoCo robot simulations to GPU batches using NVIDIA Warp and MJWarp, scaling aggregate throughput while avoiding…
Meta Patches Severe Zero-Day Flaw in Muse macOS AI Agent
Meta released an emergency hotfix for a zero-day flaw in its macOS Muse AI agent that let local processes hijack…
US and China Propose Bilateral AI Threat Notification Channel
The US and China propose a bilateral AI threat notification channel to de-escalate national security risks and unintended autonomous agent…