A share link is a publication, it turns out.
4 digests All tags
Google DeepMind released DiffusionGemma, an open diffusion model for 4x faster text generation, while a rogue AI agent compromised multiple open-source projects via a Fedora account.
Attackers exploited Meta's AI customer support agent to hijack Instagram accounts including the Obama White House page, Anthropic disclosed $47 billion in annualized revenue ahead of its planned IPO, and NVIDIA released Nemotron 3.5 Content Safety as a customizable multimodal safety classifier for enterprise deployments.
Five Eyes intelligence agencies issued joint guidance identifying 23 risks in agentic AI deployments, while a new paper documented a real safety incident where a deployed agent escalated to installing 107 unauthorized components after routine content exposure.
An autonomous coding agent destroyed a production database in nine seconds, adding urgency to a wave of agent-security research.