Overview
- Anthropic published an updated “constitution” for Claude that prioritizes comprehensive safety over ethics, policy adherence, and helpfulness in that order.
- The guidelines instruct models to resist manipulation, disclose when tasks cannot be completed, refer users to emergency services for urgent situations, and never present themselves as human.
- Anthropic states it is uncertain whether Claude has consciousness or moral standing and advises treating the system as if it may have an identity to reduce safety risks.
- Separate reporting describes a ninefold revenue increase in 2025 and estimates that Anthropic could reach a valuation above $350 billion, positioning the company closer to OpenAI in influence.
- Developers are adopting Anthropic’s Claude Code agent architecture, which can read and write files but struggles on larger projects, prompting third-party infrastructure to manage complex workflows and intensifying a three-way contest with Google and OpenAI.