Pika & AI

Early on in Pika’s existence we took a stand on AI: Pika Blogs Do Not Train AI Tools. We have not relented on this principle here at Pika, to the best of our abilities with what’s under our control. Today, the list of AI crawlers we block in the robots.txt of Pika blogs is quite a bit longer now (h/t to ai.robots.txt and others).

(We know not all AI crawlers respect the robots.txt directive. If you know of other strategies for keeping AI out of your blog on the public web, please do reach out and we’ll see what we can do to incorporate those ideas.)

In the years since that post, AI has grown to mean and do so much more than it did back then. We know some folks in the blogging community have developed even stronger stances against AI, for good reasons. We respect if you have, so we think it’s important to be transparent about our own use of AI/LLMs and how we hold the line on the principle above.

Writing code

Pika was not originally built with AI, but we, Good Enough, a small team of two, have begun using LLMs as coding assistants to help us continue to build and develop Pika the product. We’ve used them to prototype ideas, perform large repetitive coding tasks, find and close security issues, start work on new features, do first-pass code reviews, help us deal with spam and abuse, begin translations for languages we don’t speak, and other similar efforts. With practice and a great deal of patience, we’ve found their assistance so far pretty helpful, to us, in growing this personal blogging platform.

That all said, collectively we have nearly 50 years of professional coding experience, and we still live and work in our IDEs ourselves day-to-day. We both still enjoy coding and crafting software by hand! That includes Pika. We always review the code that LLMs output, and quite often rewrite much of it. Pika is not vibe-coded and neither is your blog.

Of course, to make sure Pika Blogs Do Not Train AI Tools, we do not use your writing in our development environments.

Both of us blog on Pika ourselves and our standards are quite high when it comes to the user interfaces and user experiences we want to use and deliver to customers. The design and personality of our software comes from us, humans with opinions and tastes. We hope this shines through in your use of Pika.

Content moderation

As a hosted user-generated content platform, Pika is consistently hammered by spammers, content farmers, bots, and worse. Further, we have a responsibility, both moral and legal, to do our best to create a safe environment for writers and readers of our platform. Dealing with trust and safety needs by hand as a team of two is becoming a very time-consuming challenge as Pika continues to grow.

We’ve recently begun experimenting with an internally-built moderation system to help us with this challenge. To make sure Pika Blogs Do Not Train AI Tools, we’ve built our own fully-private flagger that runs on private servers using only privately-installed open-source flagging models. This system is not accessible to any third parties, no data is stored in it, and there’s no possibility of any AI training on your writing. This system is also completely informational and can’t take any moderation actions, just flag content for us to review.

So far in our testing with this system in place, we have not only saved hours in discovering and dealing with spammers and other nonsense, we have uncovered a number of posts that threatened real harm to real people out in the real world. We’ll be continuing to monitor its usefulness.

Here’s the summary

• We do use AI to build and develop Pika.

• Pika blogs do not train AI tools, to the best of our abilities with what’s under our control:

  • We do not use your writing in developing code.
  • We do not use a moderation system that can train on your writing.
  • We do block as many AI crawlers as we can from reading your blog.

• We do not push AI code to Pika production without reviewing it (and quite often rewriting much of it).

• We do not use AI to design Pika’s user interfaces and user experiences. Those are hand-crafted by us humans.

• We do not use AI to speak for us. These are our words, as are our blog posts and other writings.

• We do not use AI for Pika customer support. We’ve previously shared our strong opinions about providing great human-to-human support, and hope the care and respect we have for our customers is apparent every time you interact with us.

We have touched on AI in various writings over the years before, but we don’t make our AI opinions a frequent topic because, truthfully, we don’t feel “all in” on the sauce here. We’re not huge cheerleaders nor are we staunch opponents. We’ve tried to understand its benefits to us as a small business, but “AI” comes with some very large caveats. In discovering where LLMs fit into our workflows, we’ve approached them with a great deal of caution and skepticism, and done our best to make thoughtful choices while balancing our curiosity and usage.

Our goal has always been to build a simple, personal blogging platform that makes it easy and enjoyable for you to express yourself on the web. We believe the ways we’ve used AI/LLMs have been effective in moving us closer to that goal.

As always, if you have any questions, comments, or concerns, you can always contact our team.

Last updated: September 2026