The Catch's AI Policy

For purposes of this policy, AI will denote a large language model (LLM), a search engine that inherently incorporates LLM models or AI-generated content such as images, music, video or other media.

The Catch's policy on AI is very simple: There are strict and limited use cases for AI. Its use in journalism should be minimized to every extent possible.

LLMs will always generate false information. Its mechanics do not have the ability to separate between the truth and falsehoods. But these services are designed to keep the user engaging with it, and are trained to fake answers they do not have. This creates an incentive to present them as "intelligent" and essential tools. They are not. AI usage has been linked to cognitive decline. LLMs and media generation models are built off of artistic theft. Its use has been recorded in multiple mass violent events. AI companies profit off of low-wage labour in Third World countries.

With that context in mind, here's what AI may be used for at The Catch:

1. Transcribing interviews and audio.
2. Services that cram AI into their interface as to make it difficult to avoid.

Results from both of these instances will be double-checked for accuracy before put into any story.

For a comparative example, the online photo editor tool Pixlr is often used to create images for stories. This services touts AI features which are not used. Meanwhile, Google search has integrated Gemini to the point of interference, and as such some research may constitute the use of AI.

That's it.

AI will not be used for generating text, media, research or source contacts. AI may only be used for a story if the subject of the piece is reporting on AI products. For example, if NotARealCompany's Chatbot "Narcissus" is giving users detailed instructions on creating pipe bombs, "Narcissus" may be used by a reporter to verify this information.

No story will be accepted if AI-generation is part of the process outside of these instances. If it is discovered that AI-generation was used as a part of the story in violation of this party, it will be clearly stated in a correction at the beginning of the piece and appropriate action will be taken.

Updates to this policy will be provided with a date stamp.