Red Teaming Methodology & Tooling
Design threat models and evaluation methods. Document them well enough for another team to reproduce the work later.
Syntony invites researchers, security specialists, and policy practitioners to join defined projects. The work covers AI evaluation and governance, along with related policy and strategic risk questions.
Syntony brings affiliates into projects when the research question needs expertise beyond the core team. The work may produce a paper, evaluation method, governance design, policy analysis, or risk forecast.
Affiliates may co-author public research or contribute to confidential client work with frontier labs, public-sector organizations, and enterprise teams. We invite affiliates based on the expertise a project requires and their availability.
The research question, responsibilities, deliverables, timeline, and authorship terms are agreed for each project.
Projects should leave behind a method that can be reused or evidence that can support later evaluation and governance work.
Where the project and confidentiality terms allow, affiliates publish under their own names with editorial support from Syntony.
Each project identifies who needs the result and what question the work must help that reader answer.
Design threat models and evaluation methods. Document them well enough for another team to reproduce the work later.
Define decision rights, escalation triggers, review requirements, and ownership for the controls that follow an evaluation.
Translate evaluation findings and risk evidence into policy analysis for public authorities and international counterparts.
Develop scenarios and capability forecasts that use evaluation evidence alongside expert judgment.
Study how synthetic media incidents can be verified and attributed, including the international humanitarian law questions they raise.
Map how deployed systems alter the behavior they measure and reshape surrounding incentives.
Applications are reviewed on a rolling basis, and we respond to each submission.