Project summary

There is no one building independent, vendor-neutral worker-side evaluation capacity of AI systems, even as they become more deeply embedded across industries and roles. I started the Center for AI Worker Safety (CAWS) to build that capacity so that workers can independently evaluate how AI is designed, developed and deployed in the workplace. This funding would allow me to go full-time from nights and weekends while key relationships are warm and preparation for a field visit is underway. My current bottleneck is hours in the day to meet the demand, rather than lack of interest. I previously spent 10 years managing international development and disaster operations, including overseeing $750M-$1B in annual emergency programs in the Middle East.

CAWS is testing whether existing workplace due-process protections and occupational safety frameworks can become practical AI governance tools once workers have the evidence and evaluation methods to use them. I’m building CAWS to answer that from workers themselves and to give them the evaluation capacity for how AI impacts their health, safety, and livelihoods. 

This funding will allow me to develop the first suite of tools for CAWS to survey worker intents and concerns around AI and develop prioritized hazard mapping tools, training, and governance resources based on worker input. The aggregated, anonymized data, analysis frameworks and governance tooling will be independent public commons goods, while the specific tools and recommendations are developed for individual unions and worker organizing groups for use in their industry. By building the cross-sector quantitative and qualitative data analysis tools, we are able to understand how workers in high-risk industries view the risks and opportunities of AI. This data and future frameworks allow workers and worker organizers to set the evaluation standards and thresholds for how AI impacts or interacts with their job. CAWS starts with workers to grapple with current harms from AI tools while building the capacity to evaluate and challenge autonomous systems and agents from being deployed in ways that are antithetical to workers’ safety and priorities. If CAWS provides workers with evaluation capacity for AI safety and health impacts, that distributed capacity is a structural check on AI development without waiting for regulators or politicians to act.

This isn't starting cold. Over the past year I've been in direct conversation with several worker-organizing bodies and a federally funded occupational-safety-training network about how AI is showing up in high-hazard work and what to do about it. I've been invited to join a federal working group that shapes how AI enters worker training, and this fall CAWS makes its public debut at a national cross-sector convening. The relationships, the ecosystem, and the first deployment are already lined up, it’s just missing hours in the day.

What are this project's goals? How will you achieve them?

To test whether workplace due-process and occupational safety frameworks hold up as real governance tools for AI once workers have the instruments and evidence to apply them.

-Open-source survey, interview guide, focus group guide, and pre-survey questionnaire

-Pilot deployment of survey, analysis, and validation tools with at least one worker-organization 

-Solicit feedback and validate survey and analysis tools directly with pilot participants

-Deliver one public workshop or presentation to a national cross-sector audience of worker organizations and occupational safety practitioners (Fall 2026).

-Publish the analysis method and one public findings report for a single sector, documenting worker-identified hazards, where due process broke down, and what workers said would work better.

-Identify research partners for questions the data surfaces but can’t answer, including how AI-assisted work affects skill retention over time and how workers interact with pace-setting and decision systems. CAWS brings the worker-validated priorities and the field access; partner researchers execute and publish.

Evaluation runs two directions here. CAWS teaches workers how to evaluate AI systems, while CAWS applies AI-assisted analysis to worker data and holds its own tools to the same transparency, notice, explanation, and appeal standards it asks workers to demand.

How will this funding be used?

Minimum: $15k (instrument development & pilot only)

• Salary, $11k 

• Field data collection and travel, $2k

• Instruments and data tooling, $1k (platform, storage, translation, software, API credits)

• Research costs, $1k (standards and literature access)

Ideal: $70k (full-time instrument/tool development, pilot + additional field testing)

• Salary, $50k 

• Field data collection and travel, $10k (2+ data collection trips)

• Instruments and data tooling, $3k (platform, storage, translation, software, API credits)

• Research costs, $4k (participant incentives, standards and literature access)

• Contingency, $3k

Who is on your team? What's your track record on similar projects?

It's just me! I have been looking for this job for the past 10 months and can't find it. I kept getting enthusiastic feedback about the idea of worker organizing as the lever to build social resilience to withstand transformative AI, by using the occupational safety/health frameworks that already exist for mapping hazards and measuring harm in the workplace. So now I’m starting it, and this funding would allow me to move from bootstrapping this organization to giving it the hours it already demands. CAWS is currently a one-person organization, and I am actively building an advisory network spanning occupational safety, labor, AI governance, evaluation, and nonprofit leadership to guide the organization’s growth.

I spent 10+ years in Asia & the Middle East working in international development and emergency response operations. I kept $750M-$1B portfolios moving in war zones/disaster stricken areas to support 2.5M people with food, water, medicine, and other life saving support every month. I spent the first part of my career working to mitigate the fallout when governance fails and humans’ lives are on the line, and now I'm working to build workers’ capacity to evaluate the AI tools of today, and by extension the capacity to meet the transformative AI that's coming. I'm used to elevating the voice of those most exposed/impacted, but least consulted to the policy makers, researchers, and corporations that make decisions about their lives and safety. Same job, new hazards.

What are the most likely causes and outcomes if this project fails?

-Budget cycles for labor unions and federal partners are slow, bureaucratic, and constrained right now. Interest and relationships are genuine but business procedures & contracting processes lag. This grant allows me to bridge the gap between quitting my job and delivering the first output. If my LOE is limited to nights/weekends, many key warm relationships and project momentum will stall, delaying development of the tools and first field deployments/analysis and subsequent refinement. Once the tools are delivered and tested with a pilot cohort of workers in a high exposure/high risk sector, downstream revenue/grant opportunities become more achievable. 

-If no one builds independent, vendor neutral capacity for workers to evaluate AI on their own terms based on their own priorities, then workers remain reliant on third party auditors and vendors to certify their own tools as safe, reliable, and worthy to be deployed in the workplace. Workers’ safety is already being harmed now, and their ability to demand workplace AI protections and resist the concentration of power risks that come with AGI/ASI, will continue to be neglected. The harms workers and society face will worsen, and workers will be injured, displaced, and forced to pay the price. Even if CAWS doesn’t get the financial demand to be sustainable, the tools produced remain publicly available for anyone to pick up and use on their own. 

-If workers do not find occupational safety and due-process frameworks useful for evaluating AI systems, that is itself an important research finding. CAWS will document where these frameworks break down and what alternative approaches workers identify as more effective.

How much money have you raised in the last 12 months, and from where?

Currently $0. I am also applying to different funding streams to complement this grant to take CAWS full time. I’m pursuing other career-transition and new-org funding, alongside ongoing discussions for potential paid engagements in varying scopes.