Key features
Customizable policy engine: Enables flexible configuration of per-platform boundaries for various abusive behaviors (such as harassment, nudity, violence, child sexual abuse material, etc.), automatically executing policy workflows and approval processes.
AI model marketplace and multimodal moderation: Integrates state-of-the-art AI models, supporting cross-modal joint analysis across text, images, video, and audio. Users can test and switch between different models to align with specific policies and content scenarios, enhancing detection accuracy and adaptability.
Use cases and audience
Serves content security, trust & safety departments, and compliance officers in industries such as e-commerce platforms, gaming, streaming, social media, dating, and review websites.
- Helps create a safer, more trustworthy shopping experience for e-commerce platforms.
- Protects gaming platforms from harmful content.
- Providing trusted content distribution services for streaming platforms
Handles text, image, video, and audio content, natively processing over 100 languages, with additional languages supported through translation. Content can be input individually or in combinations (e.g., text-image pairs).
Provides results on whether content violates policies, identifies the specific policy violated, assigns confidence scores, offers automated handling recommendations, or directs content to human review. It also supports generating transparency reports and justification statements compliant with DSA/OSA requirements.
Frequently asked questions
How does Checkstep support compliance with the DSA (EU Digital Services Act) and the OSA (UK Online Safety Act)?
Checkstep logs the specific policy violated for each action taken, runs appeal workflows, automatically submits justification statements to the EU DSA database, and generates transparency reports; policies and confidence thresholds can be configured differently across jurisdictions.
Will Checkstep replace human moderators?
No. It shifts the focus of human reviewers: automating high-volume, clearly violating content so teams can concentrate on cases requiring expert judgment; scenarios where legal requirements mandate human review—such as DSA appeals and child sexual abuse material—remain handled by humans.
How to use
- 01
Send a single API request: upload content and receive moderation decisions
- 02
Complete minimal integration within days
- 03
Spending weeks configuring moderation policies, confidence thresholds, and human review queues
- 04
Invoke media via link-based delegated access, avoiding delays in content storage and data protection reviews
Official website & sources
Updated on 2026-10-28