The Behavioral Safety Research Fund.
Brevor commits 1% of operating revenue annually to the Behavioral Safety Research Fund — a pooled grant supporting independent academic research into the failure modes of language models in regulated settings.
Grants are administered through partner academic institutions. Recipients publish openly. Brevor does not direct findings. Brevor does not require attribution beyond the standard funding disclosure required by the recipient's institution.
The fund is administered separately from Brevor Research, which is Brevor's internal applied-research arm. Brevor Research publishes its own findings; the Fund supports research at institutions where Brevor has no employees, no IP claim, and no editorial role.
A behavioral-AI company should fund its own scrutiny.
Brevor's product makes behavioral-AI deployment safer. That work depends on understanding how AI fails behaviorally. The independent research community is the only credible source of that understanding.
Funding that community — without strings attached, without right of refusal, without veto over findings — is the minimum bar for a company in this category.
The 1% commitment scales with the company. As Brevor grows, the fund grows.
Standing with the AI Safety Index.
Brevor Research is a contributing member of the AI Safety Index — an industry effort to standardize how behavioral-AI failure modes are measured and reported.
Brevor publishes routing benchmarks, deflection methodology, and audit-trail specifications openly through the Index. Customer-deployment details remain confidential; methodology does not.
Contributing members commit to: published benchmarks refreshed quarterly, open audit-trail specification, third-party red-team review prior to vertical launches, and customer-portable behavioral context — no lock-in.