Year 2+ digital twin co-design
Contribute anonymized evaluation harnesses, red-team/blue-team control protocols or production telemetry schemas for realistic lab and enterprise simulations.
Governed campus · planned launch model
The proposed campus pairs senior safety researchers and visiting industry experts with specialized instructional and red-team agents. It is designed as a reference deployment of the MCML Standard—not a replacement for human academic judgment.
Separation of powers
Legislative
Human researchers, industry experts and curricular stewards define standards, audit agent behavior and calibrate syllabi against research and regulatory frameworks.
Executive support
24/7 Socratic walkthroughs, anti-sycophancy critique and adversarial sandbox exercises under signed governance cards. Zero credential-granting authority.
Judicial & credentialing
Human faculty and visiting industry experts co-chair live, unscripted Viva Voce defenses. Final credential approval and signing authority remain human.
Year 1 AI Control Lab · Year 2+ digital twins

Replicated multi-tenant clusters, red-team/blue-team control games and model-diffing pipelines.
Planned collaboration on anonymized incident topologies, evaluation schemas and reference telemetry. No affiliation with a named frontier lab is implied.
Experiments designed to run behind active gateways with kill-switches and version-bound audit receipts.
Outcome gates & transparency
The launch plan proposes four 50-seat cohorts a year, one each quarter, subject to funding and capacity, toward a target of 200 free places a year. Growth to 100+ seats will depend on verified completion and capstone pass-rate thresholds; thresholds have not yet been approved.
Cohort 1 results not yet available.
Cohort 1 results not yet available.
Cohort 1 results not yet available.
Organization, lab & university partnerships
We invite AI safety research labs, universities, third-party evaluation organizations and enterprise security teams to help shape realistic training and benchmarking environments. Partnerships are being scoped for launch; no affiliation with a named frontier lab is implied.
Contribute anonymized evaluation harnesses, red-team/blue-team control protocols or production telemetry schemas for realistic lab and enterprise simulations.
Propose open-source runtime governance benchmarks, Agentic Governance Card compliance profiles and interpretability audit suites alongside PLEYVERSE LABS.
Sponsor cohort seats or named scholarships and explore recruitment access to learners assessed through live adversarial Viva Voce defenses.
PLEYVERSE LABS ↗ is a separate R&D entity in the PLEYVERSE ecosystem.
Submit a Lab / Enterprise Partnership InquiryAcademic partnerships & convenings
University co-ops and academic-aligned MCML and AI safety labs are proposed subject to institutional agreements; no affiliation with UH, Rice or TSU is asserted. A sponsor-funded roadmap proposes a Year 1 virtual summit, Year 2 Houston convening and Year 3 annual flagship conference.
Faculty & tutor recruitment
Help train the next generation of AI safety engineers and governance leaders. We are seeking experienced researchers and practitioners for planned paid part-time, adjunct and cohort-based roles. Appointments, compensation and launch dates will be agreed before work begins.
Lead weekly office hours, code reviews and empirical replication sprints in Compute Governance, vLLM/MCP Security, TransformerLens/nnsight Interpretability or AI Control.
Co-lead no-code Neuronpedia labs, vendor safety case audits and MCML regulatory cross-walk workshops for CISOs and risk leaders.
Join the planned human-chaired tribunal assessing 45-minute Foundational and 90-minute Comprehensive Capstone defenses. Final credential authority remains human.
Track 1 & Track 2 · companion resources in preparation