LLMs are moving from generating artifacts to creating hyper-custom worlds on demand, but still lack the ability to natively perceive and audit what they create (Andrej Karpathy/@karpathy)

For engineers, this means that as LLMs generate increasingly complex software environments, the models themselves cannot be trusted to verify the security or correctness of their output. Builders must implement external auditing and perception mechanisms to validate these hyper-custom worlds, as the generator cannot serve as its own auditor.