Weekly News
Showing all articles filed under the Weekly News category, listed newest to oldest.
Weekly Model Recap: 15 Frontier & Open-Weight Releases (2026-07-25)
Executive digest of 15 newly indexed AI models across reasoning, video generation, open-weight efficiency, cybersecurity, and quantization.
Weekly Recap: OpenAI GPT-Red, Security Research, and Agent Robustness
This week, we explore OpenAI's new GPT-Red system for automated safety red-teaming, alongside a broader roundup of security research, adversarial prompt injection defenses, and agent robustness.
Why Most Labs Keep Getting Model Alignment Wrong
As frontier models move closer to autonomous agentic workflows, traditional RLHF is hitting a wall. We explore why the next wave of model alignment will shift to synthetic debate and verifiable reasoning.
Weekly Recap: Gemma 4 & Claude 5 Rollouts
The race is heating up. Google DeepMind shipped the Gemma 4 (31B Dense) open-weights model, while Anthropic teased API previews for Claude 5 Sonnet. We break down the release schedules.