🏢 MiniMax Organization

Virtus Cybersecurity reports on a deliberately vulnerable LLM agent tested against a 22-attack corpus spanning seven OWASP Agentic Security Initiative categories. The study proposes a four-layer defen

Virtus Cybersecurity reports on a deliberately vulnerable LLM agent tested against a 22-attack corpus spanning seven OWASP Agentic Security Initiative categories. The study proposes a four-layer defen

The article examines why synchronous reinforcement learning (RL) training is inefficient at scale and how the open-source ecosystem has responded. In modern post-training, especially with long reasoni

Anthropic, often touted as a moral AI company, faces scrutiny for its involvement in military operations using its language model, Claude. This model has been linked to military actions in Venezuela a