Home About Who We Are Publications Contact

About OpenBWC

Our commitment to transparency and ethical AI in policing.

Our Mission

Accurately interpret Police BWC footage by employing open source ML, CV, & NLP technologies, resulting in identifiable behavioral patterns among police & civilians, analyzed in an interdisciplinary setting.

Our Vision

Improve policing through evidence-based recommendations sourced from BWC footage by leveraging AI in an ethical and unbiased manner.

Our Journey

A look back at our key milestones and our vision for the future of transparent policing analysis.

May 2024

Project Inception

Founded OpenBWC with a commitment to ethical AI and transparent policing analysis. Defined the mission: build AI tools that support public safety work without compromising privacy, due process, or community trust.

  • Guiding principles: Data minimization, privacy-by-design, clear audit trails, and responsible evaluation.
  • Initial focus: Visual understanding of encounters, structured extraction from narratives, and continuous evaluation pipelines.

June 2024

Prototype Development

Began development of the core analysis engine, focusing on computer vision for object and behavior detection.

  • Built the first end-to-end prototype: ingest → process → generate structured outputs.
  • Defined the roadmap: video segmentation, context understanding, and audio/text analysis.

September 2024

Initial Testing Phase

Conducted early testing with publicly available data to validate performance and fairness metrics.

  • Evaluated consistency across scenarios (indoors, outdoors, traffic stops).
  • Established reliability checks and documented system boundaries regarding claim reliability.

December 2024

Research Partnerships

Established collaborations with academic institutions to ensure rigorous oversight.

  • Formalized operational norms: secure data handling and redaction expectations.
  • Implemented "human-in-the-loop" review standards.

February 2025

Data Partnership

Secured the first partnership for anonymized BWC footage, enabling real-world testing and refinement.

  • Improved performance for real-world challenges: camera shake, dynamic lighting, and audio noise.
  • Adopted a continuous iteration cycle: test → analyze failure modes → refine → retest.
  • Built a structured dataset strategy with labeling guidance and quality checks.

May 2025

Platform Expansion

Expanded analysis capabilities to include audio transcription and sentiment analysis, moving toward multimodal understanding (video + audio + text).

  • Added transcription to support search, review, and summarization workflows.

2026 and Ahead

Public Launch & Growth

Launching OpenBWC publicly to engage the wider community and establishing it as a standalone organization for long-term governance.

  • Transparency: Publishing open-source repositories to promote reproducibility and external contribution.
  • Public Trust: Defining clear "public trust postures" regarding open-source vs. restricted data.
  • Measured Adoption: Expanding partnerships and strengthening protocols without overconfident conclusions.

Our Approach

Open Source

All our code is publicly available, allowing for community review, contribution, and transparency.

Interdisciplinary

Combining expertise in ML, criminology, data science, and civil rights to create holistic solutions.

Privacy-First

Implementing robust anonymization and data protection measures in every aspect of our work.

Community-Driven

Engaging with stakeholders, researchers, and the public to ensure our work serves real needs.

Ready to Make a Difference?

Join us in building a more transparent and accountable future for law enforcement.

Meet the Team Contact Us