Dynamic automated content compliance verification

A computer implemented method of automated compliance testing of regulated content items from a website hosted by a network computing system, the computer implemented method comprising:

• by a computing system comprising one or more hardware processors,
• responsive to a request from a user computing system to access a content presentation location of the website, receiving an identity of a regulated content item, wherein the regulated content item corresponds to an automotive vehicle, wherein the website is configured to present a plurality of regulated content items corresponding to a plurality of automotive vehicles;
• determining whether the regulated content item has been evaluated by a compliance checker for compliance with a set of constraints specified by a compliance ruleset by at least calculating a hash for the regulated content item and determining whether the hash exists within a compliance database, wherein the compliance ruleset specifies a set of criteria that evaluate compliance of regulated content items with the set of constraints, wherein the set of constraints comprises constraints that are applied to a set of variable inputs, and wherein the compliance ruleset comprises configuration parameters for configuring a set of machine learning models, and wherein regulated content items that have previously been evaluated by the compliance checker are omitted from evaluation in response to a request from the user computing system to access content presentation locations of the website associated with the regulated content items;
• and
• responsive to determining that the regulated content item has not been evaluated by the compliance checker:
• executing the compliance checker implemented using a set of machine learning models and based on the configuration parameters, wherein the configuration parameters specify a set of instructions that instruct the set of machine learning models on operations to perform with respect to the compliance ruleset and a variable input comprising the regulated content item, and wherein the set of instructions cause the accuracy of the compliance checker to satisfy an accuracy threshold;
• generating a prompt comprising the regulated content item, the configuration parameters, and the compliance ruleset;
• processing the prompt using the compliance checker, wherein the compliance checker uses the set of machine learning models to verify compliance of the regulated content item based at least in part on the compliance ruleset, and wherein the configuration parameters of the prompt instruct operation of the set of machine learning models in the processing of the regulated content item and the compliance ruleset to regulate operation of the set of machine learning models to prevent an occurrence of hallucinations to achieve the accuracy threshold by at least causing complete ingestion of the prompt prior to processing a start of the compliance ruleset;
• receiving a compliance determination dataset from the compliance checker that indicates whether the regulated content item passes one or more criteria within the compliance ruleset, wherein the compliance determination dataset comprises a number of entries that correspond to a number of criteria evaluated by the compliance checker in applying the compliance ruleset to the regulated content item;
• and
• outputting data for displaying a result of verifying the compliance of the regulated content item based at least in part on the compliance determination dataset.