Responsible technology starts with the people it affects
HRAH brings technical, legal, and human-rights perspectives to artificial intelligence safety and security. We focus on risks that can become concrete harms: impersonation, deepfake abuse, fraud, coercion, privacy loss, and unreliable high-impact automation.
Our approach connects a clearly defined technical problem with its human consequences and practical ways to reduce risk.
Research and defensive capabilities
- Threat modeling: assets, misuse scenarios, affected groups, and deployment risks.
- AI security analysis: prompt injection, unsafe tool integrations, data leakage, and governance gaps.
- Authenticity and integrity: approaches to verification and the detection of synthetic-media abuse.
- Defensive engineering: scoped prototypes, controls, evaluation workflows, and operational guidance.
What a collaboration can produce
Depending on the agreed project, outputs may include a risk assessment, a research report, prioritized mitigations, an evaluation plan, a defensive prototype, or a briefing for decision-makers.
We define access, confidentiality, reporting, and responsible handling before work begins. Security testing is performed only within authorized boundaries.
Who we work with
We welcome focused inquiries from research institutions, public-interest organizations, civil-society groups, and responsible private-sector teams. A useful introduction describes the system or problem, the people affected, existing constraints, and the intended public benefit.