INDUSTRY NEWSMonitorNEXT 12 MONTHS
Anthropic says it will watermark text generated by its AI models
TechCrunch AI
Factual evidence
What the source reports
Anthropic announced plans to implement text watermarking across its AI models, including older generations, to identify generated text.
Open sourceOneBench interpretation
Institutional assessment
Hype caution
Statistical text watermarking is easily bypassed through light paraphrasing or editing, rendering it ineffective as a standalone provenance control.