Join the field
Move into roles at AI safety research, evaluation and policy organizations.

PLEYVARSITY · Cohort 1 planning
Bring your experience in engineering, research, healthcare, policy or leadership to AI safety. Start with a free five-week foundation, then build practical skills in technical AI safety or applied governance.
PLEYVARSITY — The PLEYVERSE AI Safety and Governance Institute — is a human-governed, AI-augmented online learning institute. Its faculty is designed to pair senior human safety researchers and visiting industry experts with a network of specialized, MCML-governed AI faculty agents with separated authority inside a live governed compute lab—so every credential is earned by reproducible evidence and human-led defense, not attendance.
Home to PLEYVERSE LABS, our research lab and steward of the MCML open standard, and AI by Practice, our school for medical professionals.
Illustrative lab environment · launch infrastructure planned
200 free Track 1 places a year (target)
$0 tuition subject to grant approval · mid-career stipends
5-week front door + 12-week panels
2-year accelerated to 5-year flexible pacing
AI Control Lab in every panel
Planned MCML gateway · kill-switch drills & cryptographic receipts
Outcome-gated scaling
Planned 50-seat cohorts, four a year · verified outcomes before growth
Choose your track
GRANT-FUNDED · $0 TUITION · SELECTIVE ADMISSION
Module 00 welcomes every domain. The full hands-on program is for technical specialists in software, ML, security, infrastructure and clinical AI systems. Policy and operations learners continue into Track 2, with no coding required.
Explore track ↗
PAID BY ORGANIZATIONS · $2,500/SEAT REF.
For enterprise teams, CISOs, risk/compliance leaders, healthcare leaders (AI by Practice) and university partners—no coding required. Vendor safety auditing, no-code Neuronpedia feature inspection, AI Control containment policy and deploying MCML agent governance cards.
Explore track ↗
Every panel includes hands-on work in the AI Control Lab: real agents behind the MCML enforcement gateway, live kill-switch drills, scripted attack scenarios and tamper-evident evidence to read.
Cohort 1 pilot · planned
Year 1 offers four cohorts of 50 (200 free places). Cohort 1 is the pilot: all 50 start free with Module 00; technical specialists who pass can apply for Panel 1, and policy, risk and operations learners continue free into a 4-week governance practitioner course. We'll publish its outcomes before Cohorts 2–4 begin.
See a sample lab exercise →The certificate loop
Start with the 5-week front door, then build production skills one 12-week panel at a time. Stack evidence toward the 4-panel Foundational and 8-panel Advanced Certificates. Track 1 is planned at $0 tuition subject to grant approval, with selective admission and grant-contingent mid-career stipends.
00 ↗
5 weeks · part-time · small facilitated cohorts
Open to every domain. Learn how AI systems fail and how AI control keeps an untrusted model in bounds. Technical specialists can then apply for full Track 1 panels; policy and operations learners continue into Track 2.
01 ↗
12 weeks each · domain-grouped cohorts
Compute Governance, Zero-Trust MCP Security, Mechanistic Interpretability, AI Control and Executable MCML Governance—with continuous hands-on AI Control Lab work in every panel.
02 ↗
First milestone · 4 panels
Awarded on completing Panels 1–4 inside the AI Control Lab, capped by a 45-minute human-chaired Capstone Viva Voce.
03 ↗
Full program · 8 panels
Complete all 8 panels in 2 years accelerated or up to 5 years flexible pacing, ship an open-source governed multi-agent stack and defend it in a 90-minute Comprehensive Capstone Viva Voce.
Graduate destinations
Move into roles at AI safety research, evaluation and policy organizations.
Bring AI control and governance into your own organization.
Develop your capstone into a project at PLEYVERSE LABS.
The frontier / enterprise divide
Alignment, mechanistic interpretability, AI control and catastrophic capability thresholds investigate what happens inside a model.
Permission boundaries, runtime monitoring and regulatory reporting govern the systems around it: NIST AI RMF, ISO/IEC 42001 and the EU AI Act.
Neither layer is sufficient alone. Track 1 engineers build runtime controls; Track 2 leaders author and audit the policies. Unite both in a single defense-in-depth architecture.
Five frontier initiatives
Frontier frameworks, hardware profiling and verifiable model provenance.
Panels 1–2 ↗02Serving isolation, protocol boundaries and zero-trust MCP enforcement.
Panels 3–4 ↗03Circuit analysis, model diffing and reproducible white-box auditing.
Panels 5–6 ↗04Adversarial evaluations, trusted auditing and scalable oversight.
Panel 7 ↗05Signed MCML governance cards, runtime receipts and regulatory mappings.
Panel 8 ↗Accountability by design
The planned campus separates curricular governance, AI instructional support and human-chaired examination. AI agents cannot grant credentials.
Explore the governed campusBuild the ecosystem
LAB, ENTERPRISE & UNIVERSITY PARTNERSHIPS
We invite safety research labs, evaluation organizations and enterprise security teams to contribute anonymized control benchmarks, shape Year 2+ digital twin environments and explore a planned talent pipeline.
Partner With PLEYVARSITYFACULTY & TUTOR RECRUITMENT
Seeking researchers and practitioners in interpretability, AI Control, MCP security and enterprise governance for planned paid cohort tutoring, capstone mentoring and human-led examination roles.
Apply as a Tutor / Visiting Expert