The Conversation published an analysis of how Reuters, the BBC, and The Guardian are deploying AI in their workflows — three outlets, three regimes, three answers to the same question [1]. The paper's gap is published vs enforced policies: three newsrooms have written AI regimes, but whether those regimes are enforced, by whom, and with what consequences is a separate record.

Reuters has heavily incorporated AI to speed up its workflow, using internal tools like Lynx Insight to analyze financial data in real time and Fact Genie to process corporate press releases. The outlet can publish some breaking financial alerts in six to eight seconds [1]. The BBC has taken a cautious approach, with tools like At a Glance and BBC Style Assist used only after journalist review. The BBC published a study showing AI assistants misrepresented news content 45% of the time [1].

The Guardian's March 2026 updated policy permits limited generative AI use for tasks like suggesting alt text and transcribing audio, provided the tool aligns with editorial standards and receives senior-editor permission. The outlet's trust ownership model helps insulate it from commercial pressure [1].

WAN-IFRA's concurrent analysis asked what AI cannot replace — the answer being the editorial judgment that distinguishes journalism from content production [2]. A policy announcement is not a practice change. Published rules and enforced rules remain distinct records.

-- MAYA CALLOWAY, New York