AI Safety Intern

National Fair Housing Alliance•Washington, DC

5d•Hybrid

About The Position

Operating within the Responsible AI Lab of NFHA, the AI Safety Intern will develop original measurement and modeling frameworks for assessing the community-level risks and benefits of data center infrastructure as it pertains to artificial intelligence systems. This internship is grounded in an emerging and urgent body of scholarship demonstrating that AI systems may generate carbon emissions equivalent to a major global city and consume water on the scale of the entire global bottled water market annually and yet remain largely opaque to public scrutiny due to inadequate corporate disclosure. The Intern will work on developing quantitative frameworks that translate macro-level environmental footprint estimates into community-scale impact assessments, with particular attention to the disproportionate burdens borne by low-income communities and communities of color proximate to data center siting decisions. The AI Safety Intern will leverage AI-assisted research tools to synthesize corporate sustainability disclosures, model environmental performance metrics including Power Usage Effectiveness (PUE) and Water Usage Effectiveness (WUE) across data center portfolios, and identify disclosure gaps that obscure the true cost of AI infrastructure to affected communities. The Intern will contribute to both technical research outputs — including measurement models, data visualizations, and computational analyses — and policy-oriented deliverables aimed at informing municipal land use regulation, federal environmental disclosure requirements, and civil rights advocacy related to environmental justice. This position is designed for an emerging scholar with quantitative aptitude and a commitment to applying technical rigor in service of community protection and AI accountability. This internship position will report to the Chief AI Officer at NFHA, working full-time for a period of eight (8) weeks and expected to work in our DC Office on Pennsylvania Avenue on Mondays and Thursdays and may work remotely the remaining days of each week.

Requirements

Current enrollment in or recent completion of an undergraduate or graduate degree program in Environmental Science, Environmental Engineering, Data Science, Statistics, Geography, Public Policy, or a closely related field with demonstrated quantitative coursework; interdisciplinary candidates whose work bridges technical and social science perspectives are strongly encouraged to apply.
Demonstrated familiarity with environmental footprinting methodologies, including life cycle assessment concepts, carbon accounting frameworks (such as the Greenhouse Gas Protocol), and water stress metrics, whether acquired through coursework, independent research, or applied project work.
Evidence of research or analytical capability appropriate to career stage, such as an undergraduate thesis, capstone project, research assistantship deliverable, or independent analysis demonstrating the ability to collect, process, and interpret quantitative environmental or sociotechnical data.
Proficiency in at least one quantitative programming environment — Python or R preferred — with demonstrated ability to collect and process structured and semi-structured data from corporate sustainability reports, regulatory filings, and open government datasets, and to construct and communicate quantitative models of environmental performance metrics. Stata, Matlab, Tableau,
Working familiarity with geospatial data tools and visualization platforms (such as QGIS, ArcGIS, or Python-based libraries including GeoPandas and Folium) sufficient to map data center infrastructure against community demographic and environmental vulnerability layers using publicly available census and environmental datasets.
Practical experience using AI-assisted research tools — including large language models for literature synthesis, data extraction, and document analysis — and the capacity to apply these tools responsibly in research contexts requiring careful verification of outputs and transparent documentation of methods.
Ability to work with corporate sustainability disclosures, SEC filings, EPA emissions inventories, IEA energy statistics, and similar public data sources, including the capacity to identify and critically assess the limitations of incomplete or inconsistently reported environmental metrics of the type that de Vries-Gao’s analysis identifies as a central obstacle to accurate AI footprint estimation.
Clear and adaptable written communication skills, including the developing ability to present quantitative research findings in formats appropriate for varied audiences — from technical research memoranda for academic and scientific reviewers to community-accessible fact sheets and policy briefs for advocates and local officials.
Intellectual curiosity and scholarly rigor, including a disposition toward epistemic humility in the face of data limitations and disclosure gaps, and a commitment to representing the uncertainty inherent in environmental footprint estimates accurately rather than overstating the precision of modeled outputs.
Collaborative orientation and comfort working in a cross-disciplinary professional environment alongside legal staff, civil rights advocates, policy researchers, and senior technical leadership, with the initiative to ask substantive questions, seek feedback iteratively, and contribute actively to a shared research agenda.
Genuine commitment to environmental justice and community protection, grounded in an understanding that data center siting and AI infrastructure decisions have material consequences for real communities, and that the purpose of technical measurement work in this context is ultimately to strengthen accountability and expand the capacity of affected people to advocate for their own interests.

Nice To Haves

Prior exposure to or coursework in environmental justice, civil rights policy, or community-based research is a significant asset; experience engaging with affected communities, advocacy organizations, regulatory processes, or public interest research contexts is highly valued.

Responsibilities

Collaborate with NFHA program staff, environmental justice advocates, and community-based organizations to identify the data center siting contexts, geographic regions, and demographic communities where environmental risk modeling would have the greatest civil rights relevance and policy impact.
Support the Responsible AI Lab’s engagement with municipal and county planning agencies, environmental regulators, and utility commissions by providing technical research assistance on data center energy and water demand projections, helping to inform community comment processes, zoning reviews, and public interest interventions.
Assist the Lab in advising external civil rights partners and fair housing organizations on the practical implications of data center environmental risk assessments, translating quantitative findings on carbon intensity, water stress, and grid burden into accessible briefings for non-technical community stakeholders.
Participate in cross-sector working groups and academic-practitioner convenings focused on AI infrastructure accountability, representing the Lab’s community-centered research approach and contributing to the development of shared frameworks for measuring and communicating the distributional impacts of data center growth.
Develop and validate quantitative measurement frameworks for estimating the community-scale carbon and water footprints of data center operations and disaggregating macro-level AI infrastructure estimates to the facility, municipality, and watershed levels.
Construct probabilistic risk models that characterize the range and uncertainty of environmental impact estimates for data centers under varying disclosure scenarios, incorporating sensitivity analyses that assess how gaps in corporate reporting — particularly the absence of AI-specific workload disaggregation — propagate into uncertainty bounds on community-level exposure metrics.
Apply geospatial analysis and demographic modeling techniques to assess the distributional overlap between data center infrastructure footprints and the residential populations and environmental justice communities most exposed to those footprints.
Use AI-assisted computational tools to explore benefit-side modeling frameworks that quantify the economic, employment, and infrastructure co-benefits of data center development for host communities, enabling a structured comparative risk-benefit analysis that can inform equitable siting standards and community benefit agreement negotiations.
Use AI-assisted legal and regulatory research tools to systematically analyze the existing federal and state disclosure frameworks governing data center environmental performance — including EPA reporting requirements, SEC climate disclosure rules, and voluntary ESG frameworks — identifying the structural gaps that previous research identify as allowing major operators to omit AI-specific and indirect water consumption data from public accountability mechanisms.
Conduct comparative policy analysis of municipal, state, and international regulatory approaches to data center siting, environmental review, and community benefit requirements, assessing how different disclosure mandates and land use standards shape the availability of data needed to conduct credible community-level risk assessments of the kind the Intern is tasked with developing.
Research the application of federal and state environmental justice frameworks to data center infrastructure decisions, identifying legal and administrative pathways through which community-level risk findings produced by the Lab could be introduced into regulatory and litigation contexts.
Monitor and synthesize emerging federal and state legislative activity, regulatory rulemakings, and enforcement actions related to AI infrastructure accountability, corporate environmental disclosure, and environmental justice, maintaining a structured policy tracking resource that informs both the Lab’s research priorities and NFHA’s broader advocacy and fair housing enforcement agenda.
Author technical research memoranda and working papers documenting the measurement frameworks, modeling assumptions, data sources, and quantitative findings produced during the internship, ensuring that all outputs are reproducible, clearly reasoned, and appropriately calibrated to acknowledge the uncertainty inherent in estimates derived from incomplete corporate disclosure.
Produce accessible policy briefs, data visualizations, and community fact sheets that translate complex environmental footprint models into usable intelligence for municipal officials, community organizations, environmental justice advocates, and civil rights practitioners engaged in data center accountability work.
Maintain rigorous documentation of AI-assisted research workflows — including the tools, prompting strategies, data retrieval procedures, and verification steps used during the internship — contributing to the Lab’s developing institutional standards for responsible and auditable use of AI in environmental and civil rights research.
Present interim and final research findings in Lab meetings, internal seminars, and, where appropriate, external academic or practitioner forums, developing the capacity to communicate technical environmental modeling work clearly to interdisciplinary audiences with varying levels of quantitative familiarity.