Acceptable Use of Artificial Intelligence (AI) Policy in the Chicago Public Media Newsroom

Updated: February 13, 2026

At Chicago Public Media, we are committed to producing journalism that is accurate, ethical, and deeply human. As generative artificial intelligence (GAI) tools like ChatGPT, Google’s Gemini, and DALL-E become more available, this policy outlines how we use them responsibly, transparently, and in service of our mission. GAI is the term for artificial intelligence that generates new content like text, images, or video. GAI is trained on patterns of information in massive data sets to generate new outputs. While GAI may assist with certain tasks such as summarizing documents or analyzing data, our editorial and design content will be created and shaped by humans. This policy, which we expect to update frequently as we adapt to new technology, is intended to guide staff and inform the public about how we incorporate emerging technologies without compromising the trust our audience places in us.

1. Editorial Use of Generative AI

• Editorial content must, in general, be created by human journalists.
• GAI may be used only in the permitted use cases described below and with approved tools and license levels.
• Anything more than minor, permitted use as described below requires approval from newsroom leadership.

• Any significant use of GAI in content must be disclosed to the audience.

  • Significant use includes any GAI-generated language appearing in final content without substantial rewriting by a human journalist (or, in the case of translations, thorough review by a human), or any GAI-based data analysis that has not been verified through non-GAI means.
  • Significant GAI use in editorial content should only occur in rare use cases that serve an editorial purpose, only permitted after review by newsroom editors.
• Use of GAI in permitted cases described below requires prior supervisor approval.
• Under no circumstances should any content created by GAI be published without human review.

2. Permitted Uses of GAI

GAI may be used for tasks that are challenging but verifiable, such as:

• Summarizing documents
• Generating synonyms
• Drafting/brainstorming summary copy (e.g., for social media, headline or news alert creation)
• Mining datasets, if analysis can be verified through non-GAI means
• Transcription, with quotes and details verified through human review
• Translation during the course of reporting and research
• Translation for first-pass Spanish translations of our editorial content
• On-site translation tools for accessibility purposes
• Audio rendering of written articles for accessibility purposes
• Idea generation/brainstorming; i.e. using AI for initial concept development, mood boards, or inspiration that leads to original human creation
• Writing or drafting computer code
• Using AI-generated visual elements as components within larger, human-created works (e.g., AI textures or non-specific visual elements in a designed infographic, AI backgrounds integrated into original photography)
• Incorporating AI-generated graphic assets that are substantially modified, edited, or transformed through human creative input
• Enhancing workflow efficiency through AI-assisted tasks like background removal, color correction, or basic photo editing
• Crawling and compiling publicly available data for services such as STNG Wire and homicide/crime trackers

3. Prohibited Uses of GAI

GAI must not be used for:

• Creating finished visual content solely through prompting without substantial human transformation or integration
• Replicating or reproducing any journalist or actual person’s voice, image, or likeness, except to report on or exemplify AI capabilities
• Generating editorial content without human authorship and editorial oversight
• Generating images intended to represent real, specific people, places, buildings, or identifiable locations
• Creating visual content that could be mistaken for original photography or reporting

4. Accuracy and Accountability

• Newsroom staff are responsible for adhering to the editorial process for accuracy in all content they create, regardless of whether GAI was used in the process
• All GAI outputs must adhere to the same editorial and design standards, journalistic ethics, and accountability requirements as content produced without AI
• When there is doubt about the accuracy or originality of AI-generated material, staff must not incorporate it into the editorial process

5. Impartiality and Bias

• Journalists are responsible for identifying and correcting any bias in GAI-generated content.

  • GAI may amplify or reproduce biases in its training data and prompts; outputs must be held to the same ethical standards as human reporting.

6. Transparency with Audiences

• Disclosures must follow standards set by newsroom leadership.

• Disclosure is always required when:

  • A human voice is represented, such as digital text-to-audio conversions
  • Any content has not been reviewed by a human, such as digital language translation tools
  • GAI played a significant role in the creation of a story, beyond the uses permitted above.
• If unsure whether disclosure is warranted, staff should consult their manager or editor

7. Internal Transparency and Oversight

• Staff must inform supervisors and editors and get approval for each individual instance of GAI use or for integrating GAI into editorial workflows and be prepared to explain how it was used and the extent of human modification applied in their process.
• If staff has issues or concerns about actual or suspected GAI usage that does not comply with this policy, they should speak with their direct supervisor.
• Staff can elevate any concerns or questions directly to the AI Oversight Committee.

• An internal AI Oversight Committee has been established and meets quarterly or as required to:

  • Create and update an AI policy
  • Review new AI tools and use cases
  • Assess potential risks or ethical concerns
  • Update policy and standards as needed

8. Intellectual Property & Confidentiality

• Do not input proprietary, confidential, or unpublished information into any GAI system without prior consent, except approved platforms

  • Team-licensed GAI tools provided by CPM are not used to train public models and are configured to protect confidentiality
  • Chicago Public Media’s intellectual property includes all internal work products: reporter notes, drafts, memos, and internal documents