Poynter: How Indianapolis newsrooms are using AI — and where they draw the line

Poynter’s May 2026 examination of Indianapolis-area newsrooms is useful because it documents actual practice rather than stated policy, across organizations with different sizes, audiences, and resources. The picture that emerges is conservative adoption — with AI serving mostly as a tool for specific, bounded tasks — alongside genuine variation between outlets in how far they are willing to extend that use.

The most consistent application across organizations is transcription. WFYI Public Media uses a “person-first/person-last” method: reporters conduct the interview, run it through AI transcription, and then verify the output before it enters any story. Indiana Capital Chronicle and Mirror Indy follow comparable verification-first approaches. The common thread is that AI accelerates a low-stakes mechanical step while humans remain responsible for everything that touches the content of the journalism.

At the more expansive end, Black Indy Live uses AI for graphics, logos, and story editing, and previously deployed an AI avatar reporter. FOX59 and CBS4 have formal written policies stating that all writing and reporting must be the work of the human journalists involved and that any AI use must be disclosed to audiences. IndyStar requires transparency and human oversight. Chalkbeat prohibits AI for writing stories or altering images. Axios has a formal partnership with OpenAI for data analysis.

Taken together, the survey illustrates what is becoming a standard template for local newsrooms in 2026: AI is allowed when it supports human work on clear, verifiable tasks, and prohibited when it would substitute for human judgment on content or images. The boundaries are not identical across organizations, but the underlying logic is consistent — AI is a tool for specific functions, not a replacement for editorial decision-making.

Several lessons apply beyond local news:

The verification step is not optional. Every organization that uses AI transcription builds in a human check before the output is used. This pattern holds whether the organization has one AI policy or several.

Transparency to audiences matters for trust. Disclosure requirements appear in the formal policies of multiple outlets, not as a legal protection but as a statement of editorial standards.

Policy clarity helps when things go wrong. Newsrooms with specific, written AI policies are better positioned to respond consistently when a case falls in a gray area. Ambiguous guidelines create decision-making problems in real time.

The article was published as part of the Indianapolis Public Editor project, a Poynter-led pilot examining whether a public editor function can strengthen audience trust in local journalism.