The Tyee Has a New AI Policy

We at The Tyee have been thinking a lot about the potential for artificial intelligence to change how we work.

We’re not the only ones — practically every kind of institution has had to reckon with this new wave of tools that promises to replace or aid human labour.

To figure out whether we’ll use AI at all, and if so, how, we struck an internal committee made up of people across the organization — reporters, editors, audience development staff and business-side folks.

We started with The Tyee’s principles to guide our decision-making. We worked through all the possible use cases of AI tools that we could think of, and we debated and wrestled over where we stood as an organization.

We also thought about the value of journalism, especially non-profit, reader-supported journalism. Fundamentally, The Tyee’s mission is a deeply human endeavour. Our articles are written by people, based on newsgathering (like interviews, observation, research) conducted and verified by people. And will remain so.

We invite you to peruse our new policy, which we’ll update as needed. Here are the main points:

• We do not publish journalism that is written or generated by AI. We do not use generative AI to create or augment news images.
• We take ultimate accountability for everything that is published on our pages, including determining accuracy, sourcing and the verification of facts.
• We use and prioritize human judgment, creativity, labour and verification in our work.
• We recognize that readers value human inputs, and that readers value the human voice and style of The Tyee.
• We do not rely on answers given by generative AI to establish facts or framing for our journalistic work.
• We credit the work of others when we are building on it, referencing it, commenting on it or republishing it. We seek permission or rely on fair comment or Creative Commons standards.
• We understand that the answers put forward by AI tools are inherently laden with inaccuracy and bias and the outcome of existing power structures; they must not be taken as objective facts or reality.
• Given AI’s environmental impact, any use of AI tools will be done judiciously.
• We do not put private information or copyrighted work into tools that harvest data for commercial or other purposes.
• We do not use AI to misrepresent reality in images, audio, text or other media.
• We do not use any AI tools that obfuscate the steps it takes to create an outcome. The human user of a tool must be in charge of issuing specific instructions to a tool, and be able to follow and verify its process and output.
• While our internal policy guidelines allow for some experimentation with AI tools, we commit to fully vetting new tools and having fulsome discussions before adopting the use of any AI tool.

There are a few tools that have been around for quite a while, like speech-to-text transcription, or text-to-speech automations that we use to create audio editions of our articles. The Tyee team has been using these “narrow” AI tools for some time, and they fit within our new policy.

In some cases, an AI-generated image or passage of text may be the subject of a story. In that case, we’ll add information to the article to flag this for the reader.

To us, it comes down to this — is it possible that there are efficiencies to be found with artificial intelligence tech? Possibly. But we also have an implicit pact with our readers, that the information on our pages has been gathered and vetted using the practices of trusted journalism — direct human observation, verification, fact checking, etc.

And over the years, we’ve asked our readers to hire us to be their regional journalism team, putting in the work necessary to publish original work that you won’t find anywhere else. Well over 10,000 of you have taken us up on that, and we don’t take that support for granted.

So while we won’t completely tune out of advancements that help us do better journalism or reach more people, we’re keeping our eyes on the task at hand and keeping it real.