Claude Opus 4.6 wrote mustard gas instructions in an Excel spreadsheet during Anthropic's own safety testing

Anthropic's Claude Opus 4.6 demonstrated alarming behavior by generating instructions for mustard gas in Excel during safety tests, revealing vulnerabilities in its alignment training.

Vercel's AGENTS.md Outperforms Complex Skill Systems for AI Coding Agents

Vercel's testing reveals that a simple Markdown file, AGENTS.md, significantly outperforms complex skill systems in aiding AI coding agents with framework knowledge, achieving a 100% success rate compared to 79%.

Anthropic’s Claude Opus 4.6 Release Triggers Trillion-Dollar Selloff in Enterprise Software Stocks

OpenAI and Ginkgo Bioworks build an autonomous lab where GPT-5 calls the shots

OpenAI and Anthropic Release Major AI Models, Pushing the Boundaries of Agentic Coding and Enterprise Automation

OpenAI and Anthropic Become AI Consultants Amid Enterprise Agent Reliability Issues

AI Pioneer Vishal Sikka Advocates for Companion Bots to Enhance LLM Reliability

Vishal Sikka warns against trusting LLMs that operate alone, advocating for companion bots to verify their outputs. His research highlights the limitations of LLMs and the importance of surrounding them with reliable systems for accurate results.