Frameworks for safe software development
Open evaluation tools for AI safety, privacy architecture, and responsible release decisions.
Most AI safety features are configuration options. They can be disabled. At scale, they will be. If your constraints are optional, they are not constraints.
We build evaluation frameworks and the protocols they test. Open tools that help developers, researchers, and organisations ask the right questions before they ship. Open standards for the next generation of safe software, built alongside the infrastructure that implements them.
Safety that is built in, not bolted on. Infrastructure you can trust because the architecture does the work.
The tools
AI Safety Evaluation Tool
Evaluate any AI system against 7 constraint-mandatory principles. Quick check, procurement review, code scanner, and architecture review modes.
Release Decision Framework
Seven questions to ask before you open source anything that could be misused. Stripping potential, dual-use risk, licensing strategy.
AI Behavioural Testing
Test AI systems for confabulation, contradiction, gaslighting, and manipulation. Copy-paste test prompts included.
AI Psychosis
How AI affects you. Mental health safety testing and a self-assessment for your own relationship with AI.
The Slop Test
Seven questions to find out if your AI idea is ready, or if you have more work to do.
Cross-Jurisdiction Compliance
Compare AI regulations across the EU, US, UK, Australia, and China. Filter by product type.
Readiness Assessment
See where your organisation is today and what structural safety would change. Sector-specific results.
From the blog
What You Unlock When Safety Is Structural
Safety features that can be disabled are liabilities, not assets. Here is what changes when the architecture does the work.
No Offline Mode
Good systems design means never having a single point of failure. So why are we eliminating every analog fallback and moving to digital only?
NOEVA builds open infrastructure for a humane internet. Protocols, tools, and standards where safety is structural, not optional.
Want to use the tools, contribute, or just talk about AI safety? Get involved. See what safer infrastructure unlocks for your sector.
Get in touch →