decipon.com

What It Is

Decipon is an AI content analysis tool operated by Norway-based Synapti AS, built around its open standard, the Influence Tactics Protocol (ITP). It does not determine whether content is true or false. Instead, it visualizes influence tactics found in articles, posts, tweets, or transcripts, produces a 0–100 Influence Tactics Score, and shows evidence for each category detected. Its official positioning is clear: it is not a fact-checker or a truth oracle, but a “tactics lens.”

Core Capabilities and Output

Decipon breaks influence tactics into 20 categories, grouped into five composite dimensions: emotional manipulation, suspicious timing, coordinated messaging, tribal polarization, and omitted information. It can identify patterns such as fear, anger, urgency, false dilemmas, group conflict, selective data, and logical fallacies. Technically, it uses a hybrid pipeline: 15 categories are handled by deterministic ML models, 5 by a temperature-0 LLM layer, combined with rule-based heuristics. Reports provide not only a score, but also three perspectives — Critical, Supportive, and Synthesis — to avoid turning the score itself into a direct conclusion.

Pricing and Integration

The free tier allows 3 analyses per day without an account, but only offers ML-only Quick mode. Paid credit packs range from $10/50 analyses to $100/800 analyses, and credits do not expire. API/publisher plans are billed monthly at $0.03/call and paid via Stripe. Developers can use the REST API, OpenAPI, and X-API-Key authentication, and track remaining quota through response headers or the balance endpoint. Publishers can also use embedded badges, domain verification, badge customization, and analytics dashboards.

Pros, Cons, and Limitations

Its strengths are a relatively transparent methodology, with ITP, model cards, and specifications publicly available, as well as a clear warning that “a high score does not mean false.” This makes it better suited for media literacy and research support. The limitations are also significant: Decipon officially acknowledges that it does not yet have peer-reviewed validation, independent third-party benchmarks, or expert agreement studies. The SemEval benchmark only covers a mappable subset and should not be treated as validation of the full 20-category capability. On privacy, Decipon says it does not store personal information and that Stripe handles payment emails, but it does not provide detailed explanations about retention or training use of analyzed content.

Best Fit and Access from China

It is suitable for journalists, researchers, publishers, developers, and readers who want to identify signs of manipulation before sharing content. It is not suitable for people who want to treat an AI score as a ruling on truth or falsehood. Chinese-language support is not clearly stated, and there is no visible validation information for Chinese-language data. Network accessibility and payment availability in mainland China are not disclosed in the main documentation and need to be tested directly. If used for Chinese public opinion monitoring or domestic media analysis, users should first validate output quality on a small sample and prepare alternative media monitoring or fact-checking tools as backups.