Anthropic's project, known as Glasswing or involving its Mythos AI, has highlighted the security paradox of artificial intelligence. The project has reportedly identified long-standing vulnerabilities, but its own security has been called into question, with some outlets suggesting that outsiders accessed the system shortly after its launch. The extent of the project's expansion and access to it are unclear, with some reports indicating it has been made available to over 150 organizations, while others suggest that global access has been restricted.