Fail Safe: Why Anthropic won't release its new AI model

Anthropic says its new AI model is so good, it's not safe to give users access to it. If that's true it could upend the tech industry, and reframe its ongoing row with the US government, writes Adam Maguire.