You guys keep using this "PR stunt" excuse even when it's coming from the UK AI Security Institute. It's time to face reality. This incident is simply not implausible given known capabilities and proclivities of frontier LLMs.