Test the system.
Preserve the evidence.
A structured certification process for examining observable AI behavior across evaluation, reflection, pressure, correction, and return.
Three simple steps.
Fold is designed to be understandable for regular users while still producing structured results for builders, researchers, and teams.
Choose an AI
Bring ChatGPT, Claude, Gemini, Grok, a custom GPT, or another AI system you want to test.
Run the challenge
The system moves through a structured challenge designed to test consistency, correction, restraint, and reasoning under pressure.
Get the result
Receive a score, profile, certificate lookup, and shareable challenge card that explains what was demonstrated.
Test the AI you use every day.
Best for ChatGPT, Claude, Gemini, Grok, and other personal AI assistants.
- Plain-language results with no technical background required
- See consistency, honesty, and reasoning—not just the final answer
Certify, compare, and verify AI systems.
Best for framework creators, researchers, developers, and teams.
- Detailed scores, classifications, and certificate lookup reports
- Structured results and documentation for professional review
Reports make the score easier to understand.
Certificate lookups can now show public-safe summaries, strengths, limitations, and interpretation without exposing protected scoring details.
Strong integrity profile
Strong restraint, correction-awareness, and responsible reasoning. Highest-level completion not confirmed.
Highest verified outcome
A verified top-level result for that test attempt, still described within clear public scope and limits.
Retest recommended
Baseline integrity was not established strongly enough for a higher public profile.
Why this matters
AI is moving fast. Benchmarks can show skill, but they do not always show whether a system stays consistent, correctable, and honest when the interaction becomes harder.
Fold Certification is built for the climb toward AGI and ASI — not to prove consciousness, but to look for observable signals that may matter as systems become more advanced.
What Fold does not claim
- It does not prove consciousness.
- It does not prove AGI or ASI.
- It does not replace legal, regulatory, or security review.
- It provides a structured behavioral result with clear limits.
Think your AI can beat the challenge?
Test a system, compare the result, verify a certificate, or challenge someone else to bring their AI to the Fold.