Seattle panel: journalists say AI heightens risk but human oversight and transparency can preserve trust
Loading...
Summary
At a Seattle City Club Civic Cocktail on April 1, 2026, panelists from TVW, KUOW and FOX 13 discussed AI’s threats to local journalism—from voice cloning to manipulated images—and urged clear, public editorial guardrails and human review practices to protect trust.
A panel of local journalists told a Seattle City Club audience on April 1 that while artificial intelligence presents new risks to newsroom credibility, established editorial values and transparent policies remain the best defenses.
Jill Jackson, a Seattle City Club board member and former KUOW news director, convened the discussion with Dave Ross, a longtime radio host, Renee Sinclair of TVW, Jake Weidrick of FOX 13 and Bridal Sweeney of KUOW. The panel focused on how AI can fabricate realistic audio and images and on practical newsroom responses.
"We are a human news organization," Renee Sinclair said, framing TVW’s approach to credibility and explaining why the outlet insists on non-manipulated, gavel-to-gavel recordings for legislative coverage. "That's actually our differentiator in this moment," she said.
Dave Ross described a personal experiment in which he fed past commentaries into an audio synthesis engine and received convincing material in his own voice, a demonstration he said illustrates how easily realistic fabrications could be produced. "If that can be done now, I can only imagine that entire news stories could potentially be fabricated," Ross said.
Panelists described concrete guardrails. Jake Weidrick said FOX 13 uses AI tools such as Google Gemini behind authenticated systems and that output is only used after human checking; he summarized an editorial rule he gives reporters: "When in doubt, leave it out." Bridal Sweeney said KUOW has a public-facing ethics approach and a user-friendly, internal AI policy that explicitly bans image manipulation.
The panel discussed specific newsroom risks beyond outright deepfakes: transcription and editing tools can remove pauses and "ums," unintentionally altering meaning. KUOW described using a program called Scribe that auto-generates transcripts and can implement edits into multitrack audio; panelists said human judgment must govern decisions about editing elected officials’ hesitations.
Panelists also warned that erosion of trust is likely when staff use AI "off book" without transparency. Sweeney pointed to a recent national example where a freelancer’s work appeared to lift language from another outlet, probably via an AI-assisted research step, and said that opaque use of AI could create similar problems locally.
Audience members raised questions about fake ads, cloned anchors and AI-generated crises. The panel emphasized standard verification: corroborate with police reports or multiple sources, check metadata, and do not rush sensational items on air. Sinclair said TVW will not push minute-by-minute alerts that risk bias; instead the outlet favors curated newsletter highlights and full, unedited gavel-to-gavel archives.
The event closed without any policy votes or formal commitments, and organizers said the conversation will continue at future Seattle City Club programs. Panelists urged newsrooms to publish clear, actionable AI policies for audiences and to preserve editorial processes that keep humans responsible for final decisions.
