ContentMod

What is ContentMod?

ContentMod is an AI-based content moderation service designed for developers and platform management teams; it is used to analyze text and images submitted by users and to incorporate risk assessments into the product’s workflow. It is suitable for social communities, e-commerce platforms, educational tools, corporate collaboration tools, and other scenarios that involve user-generated content.

The product is centered around managed APIs and a management dashboard, and it also provides a TypeScript SDK, OpenAPI specifications, an approval queue, webhooks, as well as Discord bots. It is not a general-purpose content management system, nor can it replace the platform’s own community rules or manual appeal processes.

Main functions

Text review

• After entering a text, the interface returns a security assessment, confidence level, sentiment, theme, risk score, and sensitive category.
• The results include actions suggesting rejection or referral for manual review, and can also output a filtered text with inappropriate words replaced by asterisks.
• Custom metadata such as user IDs can be included, facilitating the association of audit records with content or accounts in the business system.

Image review

• Images can be submitted using a valid image URL or a Base64 string; they are suitable for use as profile pictures, product images, captions for posts, and attachments in chats.
• The output includes security assessments, image descriptions, risk scores, themes, content ratings, sensitive categories, and severity levels.
• Image review consumes 3 tokens per image, which results in higher costs compared to text review that uses 1 token per review.

Review queue

• The manual queue allows reviewers to accept or reject each piece of content individually, and it enables AI to generate reports first to assist in making decisions.
• Automated queues can, based on the settings, accept requests directly, reject them, or route them for manual review; they are suitable for content streams that are frequent but follow relatively stable patterns.
• The manual queue is available to all plans and can be used for free, while the fully automated queue requires a paid plan.

Asynchronous processing and webhooks

• Requests for text and images can be processed at a later time; the interface first returns an approval number, after which the results can be obtained by using a query interface.
• After configuring the callback, the service will send the completion result to the business system, thereby reducing the need for the client to perform continuous polling.
• Actions such as acceptance or rejection in the queue can also trigger webhooks, enabling automatic hiding of content, notifying users, or creating tickets.

Custom rules, glossaries, and analysis

• The platform allows for the customization of review parameters and the definition of prohibited words, enabling the same detection capabilities to be adapted to different communities or age-group requirements.
• The management backend provides a Playground, where you can use sample content to observe the output results before deciding on the thresholds and handling logic.
• Usage analysis is used to track audit calls and resource consumption; for multiple projects, separate billing is applied, which facilitates cost calculation by customer or business unit.

Discord bots

• Administrators can add the ContentMod bot to a Discord server and then link the ContentMod account using connection commands.
• Robots are used to automatically review server messages; after unbinding them, it is necessary to uninstall or remove those robots from Discord in order to completely stop access.
• When granting permissions, it is necessary to check the server rights requested by the robot, and to revoke any unnecessary permissions and project keys once the robot is disabled.

Input and output

Access process

1. Create accounts and projects, obtain the public key and private key in the project’s backend, and store them separately for development, testing, and production environments.
2. Choose the REST API or install the TypeScript SDK, initialize the client, and ensure that keys are stored only on the server side.
3. First, use representative text and images in the Playground to test the risk categories, confidence levels, and recommended actions.
4. Place the review call within the process before content is published or after it has been uploaded, and choose between synchronous feedback, delayed processing, or a webhook based on the business risks involved.
5. Create manual or automated review queues, and set rules for rejection, further review, notification, and appeal, in order to avoid relying solely on a single score.
6. Monitor tokens, false positives, false negatives, and processing delays, and regularly re-calibrate the rules using real samples.

API and SDK

Price packages

Charging is done on a project basis and processed through Stripe; separate settlements are applied to different projects within the same account. The public price is given in terms of a monthly fee in US dollars, while taxes, the currency used for settlement, and payment methods may vary depending on the actual settlement page.

Both Starter and Pro versions include image review, a review queue, multi-language support, and webhooks. The terms of service do not specify anything regarding refunds, the timing of cancellation, or what to do with unused tokens; it is necessary to check this information on the payment page or through customer support before making a purchase.

Multilingual and platform support

• The product page states that text verification covers more than 50 languages, but it does not disclose the accuracy rates for each language, any regional differences, or a complete list of the languages supported.
• The main platform for operations is the web-based management backend, while the actual review functionality is integrated into various applications through server-side APIs, SDKs, webhooks, and Discord bots.
• No iOS, Android, or native desktop applications designed for ordinary consumers were found.

Open source and commercial use

The ContentMod hosting platform and its moderation model are not considered to be open-source products. The npm package metadata for the official TypeScript SDK indicates a MIT license, but there is no separate license file in the code repository; therefore, it is advisable to verify the applicable licensing terms when reusing the SDK’s source code.

The paid plan is intended for integration with commercial platforms, but users must still ensure that the content submitted, as well as the rules for review and the final actions taken, comply with local laws, contracts, and the policies of their own communities. The outputs generated by the model do not transfer any copyright rights to third-party content, nor do they exempt users from the need for manual review and the possibility of filing appeals.

Privacy and security

• The privacy policy states that account information, API usage records, and cookies are collected, and they are used to provide services, handle billing matters, ensure security, and conduct performance analysis.
• The platform makes use of Cloudflare’s API infrastructure and security services; the front-end is hosted on Vercel, and payments are processed by Stripe on a project-by-project basis.
• The service guarantees that personal information will not be sold or rented out, but necessary data may be shared with service providers or agencies that have legitimate requests under the law.
• The policy allows users to request access to, update, or delete account data, as well as to disable unnecessary analysis cookies.
• The public documents do not specify the exact retention period for the texts and images submitted for review, the location where the data is stored, or whether it is used for model training; such matters must be confirmed through a contract before personal information is processed.
• Any automatic review may lead to errors or omissions; in high-risk scenarios, mechanisms for manual verification, appeals, logging, and failure mitigation should be in place.

Frequently Asked Questions

What types of content can ContentMod review?

It primarily handles text and images, providing safety assessments, risk scores, sensitive category classifications, and recommended actions; it does not yet offer capabilities for reviewing audio or video files.

How are tokens calculated for text and images?

Each text review consumes 1 token, while each image review consumes 3 tokens. Calls to AI from the queue should also be taken into account when planning the project’s usage.

Is there a free option?

The manual review queue can be used for free, but the official pricing page does not list all the free API packages available. The Starter and Pro plans cost $20 and $69 per month, respectively.

Can the result be used directly for automatic deletion?

Technically, it is possible to handle such tasks through automatic queues and webhook triggers, but for important accounts or high-risk content, manual review and appeal processes should still be available.

Are APIs and SDKs provided?

The platform offers REST APIs, OpenAPI specifications, and a TypeScript SDK; it is also possible to access server verification through Discord bots.

Is ContentMod an open-source product?

The hosting platform and the review model are not considered to be open source. The TypeScript SDK package is licensed under MIT, but the subscription service associated with the product does not thereby acquire an open source status.