A Look Inside the AI Strategies at the New York Times and the Washington Post (www.niemanlab.org)

At Digiday’s recent summit, The New York Times and The Washington Post revealed contrasting but complementary AI playbooks that show how legacy newsrooms are operationalizing models for both journalism and revenue. The New York Times is focused on investigative and research use cases — teams work one-on-one with reporters to apply AI (e.g., OCR/multilingual transcription of messy iPhone photos) and then formalize repeatable workflows for the wider newsroom. They pair experimentation with heavy editorial and legal caution, and maintain an open Slack channel to share tools, access (Gemini, enterprise ChatGPT) and use cases across bureaus. The Washington Post, led by newly named chief AI officer Sam Han, emphasizes revenue optimization through machine learning: an AI-driven, individualized paywall that adapts to reader behavior has increased customer lifetime value by ~20%. The system replaces manual rule-making with model-driven decisions and folds in flexible products (weekly/daily passes, pay-per-article) to test monetization strategies. Both outlets highlight governance: sensitive, business-critical tasks must run on internally hosted models while less-sensitive work can use vetted enterprise LLMs, contingent on proper document classification. For the AI/ML community, these examples illustrate pragmatic production deployments — personalization, policy-driven model selection, and turn-key workflows — offering lessons in model integration, A/B testing, and governance at scale.