Two Geko Productions
AI and data use
This disclosure applies to the TGPS platform that powers STNR.TV and Cruces Public Access.
How our software uses AI
We do not run AI models in the core TGPS software. We build and operate core features such as accounts, publishing and our own playback. Some integrated providers use AI in their own services, as described below. Some artwork, promotional material or other content you see may have been made with AI. Our developers may also use AI to help build features and troubleshoot bugs.
Customer data and development
The AI tools we use for development do not have direct access to live customer accounts or production databases. When we use AI for development or troubleshooting, we may share relevant details from a support report or technical diagnostic for that specific task. That use is limited to the information needed to develop the feature or investigate the issue.
AI used by integrated partners
Some features use outside services for video, GIFs, shopping, payments, hosting or TV distribution. Which services apply depends on the network and the feature you use. Partners can process the information they receive for those services under their own policies. The examples below describe what each provider publicly documents and how it relates to TGPS.
- YouTube · Video
- YouTube uses machine learning to generate automatic captions for eligible videos and uses viewer signals to personalize recommendations. These features may appear when you play a YouTube video or open YouTube. YouTube automatic captions · YouTube recommendations
- GIPHY · GIF search and stickers
- When you use the GIF picker, GIPHY receives the search term and returns matching media. GIPHY documents using machine learning alongside human review to moderate its library. Its developer documentation describes a search algorithm and gives no specifics about whether that API ranks results with machine learning. GIPHY Search API · GIPHY moderation
- Printful · Merchandise fulfillment
- For eligible orders, Printful receives product and delivery details to make and ship the merchandise. Its own design tools can use AI to improve image resolution and generate product mockups. Those tools are available in Printful’s Design Maker. Our integration uses Printful for catalog and fulfillment. Printful Smart Image Tool · Printful AI mockups
- Shippo · Shipping rates
- When a checkout uses Shippo, we send shipping addresses and package details through its Shipping API to get carrier rates. Shippo also offers a separate AI assistant connection for actions such as getting rates or creating labels. TGPS uses the Shipping API for rate quotes. Shippo’s Shipping API and AI assistant connection
- PayPal · Payments
- PayPal’s privacy statement describes automated decisions for fraud, risk and creditworthiness, where applicable, and describes how PayPal uses AI tools. When you choose PayPal, its safeguards can affect whether a payment is approved or reviewed. PayPal Privacy Statement
- Stripe · Payments and creator payouts
- Stripe Radar uses machine learning to assess payment risk. Stripe’s fraud checks can affect how a transaction is handled when Stripe processes the payment. Stripe Radar risk evaluation
- Amazon Web Services · Cloud hosting
- AWS hosts and delivers parts of TGPS through services such as Amazon S3 and CloudFront. AWS also offers separate AI services such as Amazon Bedrock. The TGPS hosting and delivery paths covered here use AWS infrastructure to store and serve content. Amazon CloudFront · Amazon Bedrock
- Roku · Connected TV distribution
- Roku distributes both networks’ TV apps. Roku says it uses machine learning to personalize content discovery on its platform and offers voice search on supported devices. These features operate on the Roku platform. Roku content personalization · Roku voice search
Cloud hosting and outages
An outage of the AI tools we use for development would leave the live TGPS platform running. A partner outage could interrupt the specific service that partner provides. Cloud hosting is a separate dependency. If our hosting data centers went offline, the hosted websites, streaming and account services could be interrupted, along with our ability to deploy live updates. Restoring those services would require working hosting infrastructure.