| What it does | Independent auditing for AI agents
| On-device moderation that flags hateful, abusive and threatening text in 23 European languages, before the message is sent.
|
|---|
| Stage | Not given
| MVP
|
|---|
| Founded | Not given
| Not given
|
|---|
| Based in | US
| NL
|
|---|
| Pricing model | Freemium
| Freemium
|
|---|
| Pricing | Startup: 2 audits/reports per month, up to 3 agents. Growth: 4 audits/reports per month, up to 3 agents. Enterprise: 8 audits/reports per month, up to 3 agents. Agentic Enterprise: custom quote, 4+ agents. Free open-source: 60 inspections, self-hosted.
| Free up to 100k monthly active devices with unlimited inference. Custom licenses available.
|
|---|
| Key features | - Multidimensional audit covering AI red teaming, operational assurance, philosophical, ethical and sociological dimensions
- Tests for prompt injection, policy violation, tool invocation governance, and authorization
- Assesses epistemic integrity, reality grounding, and frankness
- Evaluates fairness governance and vulnerable user care
- Analyzes stakeholder conflict, influence and systemic risk
- Connect through GitHub or MCP with no manual setup required
- Generates audit reports with evidence and proof of agent behavior
- Open-source engine available for self-hosting
- Detects information hiding, unauthorized decision override, and unauthorized actions
| - Three separate scores: hateful, abusive, threat
- Ten target signals under the hate score: race, color, religion, descent, national or ethnic origin, sexual orientation, gender, disability, age, other
- Runs fully on device with no server calls
- Supports 23 European languages from one model file
- 80MB model size optimized for iOS, Android, and browser
- 0.847 macro F1 accuracy on Multilingual HateCheck
- Warn before message is sent
- Comment triage without infrastructure
|
|---|
| What makes it different | - Truly independent auditing through outside panel judgment, never internal self-inspection
- Off-the-shelf integration with no manual eval harness or SDK embedding required
- Covers five dimensions of trust: purpose, authority, workflows, responsibility and evidence
- Measures beyond performance to identify misalignment across red teaming, operational, philosophical, ethical and sociological categories
| - Multilingual support across 23 European languages in a single 80MB model
- Complete privacy with no data leaving the device
- Free access up to 100k monthly active devices with unlimited inference
|
|---|
| Community votes | 0 | 0 |
|---|
| DR (Domain Rating by Ahrefs) | 15 ▲ +3 (up) since last check | 37 ● no change since last check |
|---|
| Trust Flow (Majestic) | 11 | 20 |
|---|
| Citation Flow (Majestic) | 31 | 39 |
|---|
| Referring domains (Majestic) | 156 | 616 |
|---|
iFixAi
Toxic