AI-text detectors fail on LLM-rewritten human writing, new benchmark shows

Original: ARB: A Matched Authorship-Rewriting Benchmark Dataset for AI-Text Detector Evaluation

Source: arxiv.org ↗