AI IntelligenceAug 22, 2026AI Intelligence
Article
LLMs could write like humans but post-training guardrails make their text detectable
Pangram CTO Bradley Emi argues that post-training and safety guardrails, not limited model capability, make LLM text detectable. Base models without those constraints already write with far more variety, he says.
Frontier EditorialSource: The Decoder
01
Source Brief
LLMs could write like humans but post-training guardrails make their text detectable: Pangram CTO Bradley Emi argues that post-training and safety guardrails, not limited model capability, make LLM text detectable. Base models without those constraints already write with far more variety, he says.
02