AI companies' DNS: 17 Anthropic-verified, 23% have spoofable email (domainintel.vercel.app)

A recent analysis of 39 AI-relevant domains revealed that 17 have verified their ownership with Anthropic, indicating participation in the Claude API enterprise onboarding process. This finding highlights the interconnectedness of AI companies, as many, like Hugging Face and Stripe, are linked through both Anthropic's and Cursor's verification systems. Interestingly, several of these domains possess DNS records formatted for the Model Context Protocol (MCP), which suggests an emerging standard for identity verification among AI services; however, the absence of a public specification raises questions about its implementation. On the downside, the analysis uncovered that 23% of the domains lack robust email authentication protocols. This includes notable AI safety organizations like the Alignment Forum and MIRI, potentially exposing them to phishing attacks. In contrast, companies like OpenAI and Anthropic maintain strict email policies, underscoring a divide in security measures among AI stakeholders. This situation demands attention as the AI community continues to navigate the complexities of security in an increasingly interconnected digital landscape.