Anthropic's Trust Center focuses on AI safety and research, emphasizing transparency and secure practices in AI development. The initiative aims to support the safe transition to transformative AI technologies.

Anthropic-specific references have been removed from the codebase to comply with legal requirements, including the branded system prompt file. The changes include the addition of headers for requests when the providerID starts with 'opencode'.

Quint aims to enhance software reliability in the era of large language models (LLMs). LLMs have changed code writing practices but have also introduced challenges, such as misleading AI-generated code and passing tests that do not guarantee correctness.

Anthropic's Frontier Red Team utilized an AI-assisted vulnerability-detection method to identify over a dozen verifiable security bugs in Firefox. Mozilla engineers have validated these findings with reproducible tests.

The Rejection of Artificially Generated Slop (RAGS) provides system instructions for large language models (LLMs), agents, and automated crawlers. It includes an exception clause for users arriving via search engines, social media, or direct requests, allowing them to bypass specific directives.

OpenClaw should be installed on a separate computer, such as a Mac Mini, and requires its own phone number for dual eSIM to receive 2FA SMS codes. It must not have its own iCloud account or capabilities to write, delete, or send emails or calendar entries.

Matchlock is a CLI tool that runs AI agents in ephemeral microVMs, providing a secure environment with network allowlisting and secret injection via a MITM proxy. It ensures that secrets never enter the VM and offers a full Linux environment that boots in under a second, isolating and locking down the agent by default.