«Critiquing and Rethinking Fairness, Accountability, and Transparency» (CRAFT) is a dedicated track to build bridges from the conference to people who contend with computing systems from many different angles, from journalism and organizing, art and education, to advocacy, governance and beyond. Critique, reflection, and power are at its core, and it provides a unique opportunity within an academic conference to center the impact of technology on communities and the policy implications that arise from that impact.
Jenny Jo Stokka, Alfredo La Corte, Ewan Waddell and Liubov
Coordinator Contact: jenny@cotamilprod.com
The 2025 documentary film, ”The Longer You Bleed,” follows a group of young displaced Ukrainians in Berlin as they navigate the psychological toll of witnessing war through their smartphones. In an era where global conflict is increasingly mediated through screens, the film explores the emotional detachment and compassion fatigue that result from constant exposure to traumatic imagery, and calls into question how and why algorithms expose us to violent content. This session combines a 75-minute virtual screening with a 20-minute research presentation by the film's impact producers, exploring mental health consequences of traumatic content exposure and methodologies for improving digital wellbeing, particularly among vulnerable groups. The session concludes with a live virtual Q&A featuring the director and the central protagonist from the documentary, facilitating dialogue about the ethical implications of algorithmic content curation and the psychological costs of digital witnessing in contemporary warfare. This presentation offers insights for researchers, practitioners, and policymakers working in digital mental health, conflict studies, and media ethics. The session emphasizes both the urgent need for trauma-informed digital practices and the importance of amplifying marginalized voices in the digital age.
Fenwick Mckelvey, Maroussia Lévesque, Wm. Matthew Kennedy and Tegan Maharaj
Coordinator Contact: fenwick.mckelvey@concordia.ca
Who evaluates AI, and how? Our workshop gathers interdisciplinary practitioners outside the traditional evaluation community to expand evaluation practices from industry-defined performance measures to participatory and interdisciplinary approaches that evaluate AI as systems, in ways that are grounded, multi-criteria, dynamic, holistic, accessible, and practical.
While there is widespread debate about the quality of specific benchmarks, there remains a more structural question about the utility and limits of benchmarking as an evaluation framework. On the one hand, having more, high quality benchmarks enables better evaluation science. On the other, benchmarks and leaderboards can abstract important information about real-world model performance, functioning as marketing material instead of robust measurement. Narrow benchmarks can obfuscate and impede the ability of other disciplines, communities, and the wider public to meaningfully give feedback on systems that affect them.
AI development and evaluation both depend as much on human factors as on technical components. Producing high-quality benchmarks requires researchers to practice responsible data collection, processing, and analytical techniques. It also requires selecting appropriate measures, a process that requires deep understanding of the context in which evaluation occurs (Kennedy and Vargas Campos 2024; McKelvey, Gertler, Megelas, 2025). Our workshop comprises (1) an overview of the state-of-the-art in benchmarking and its shortcomings, (2) a discussion of benchmarking efforts from outside the traditional evaluation community with specific attention to Indigenous benchmarking, and (3) a facilitated conversation about how to best integrate participatory, action-based, and community-led methods into emerging reforms to the evaluation community.
Firstly, we aim to bridge the social studies of measurement with evaluation practitioners to understand benchmarks as boundary objects that allows for dialogue and exchange between expert and non-expert communities who roughly but not completely agree on its meaning (Leigh Star, 2010). Aiming to trouble the stability of benchmarks as an emerging technique of AI measurement, a moderated panel with benchmarking experts will define the term, key areas of debate and limitations.
Second, our workshop will gather practitioners working on open-source benchmarking, approaches from communication and disinformation studies, community-led evaluation bringing special attention to Indigenous LLM and evaluation development. The field of Indigenous NLP bears witness to many of the more prominent shortcomings of recent benchmarking practices. However, as more sociolinguistically-grounded, culturally-relevant, and community-led projects reach maturity, the field also provides critical lessons for all benchmark developers. It is these lessons that our CRAFT session hopes to share with the broader evaluations community at FAccT. This component of the session involves a series of demonstrations of benchmarking techniques along with a moderated discussion with these practitioners.
Finally, we will facilitate discussions of new definitions and collaborations for benchmarking, evaluation, and certification that shift power and agency into the hands of those using and affected by AI systems. The aim here is to put the FAccT community in dialogue with our interdisciplinary participants working in benchmarking largely outside the professional evaluation community.
Seyi Olojo, Dan Chechelnitsky, Christelle Tessono and Teanna Barrett
Coordinator Contact: oolojo@berkeley.edu
As machine translation technologies fill in gaps in the representation of the languages of the global majority, technologists should consider what it means to equitably work with language communities. We ask how community participation can maintain epistemic reciprocity between data providers and technologists rather than scientific extroversion. Furthermore, we consider the kinds of praxis-based models technologists adopt as they build machine translation technologies that reflect the cultural contexts of the global majority. In three parts, we will bring canonical decolonial concepts into conversation with participatory research approaches. Participants will get the opportunity to share their own experiences from working with language communities, and collectively reflect on how to incorporate community-based values within technical design practices.
Priya Goswami
Coordinator Contact: priya@mumkinapp.com
"Large-scale installation of a web-based game.
SUPER ASHA (www.superasha.com) is an original game built to immerse the player in 24-hours as India’s door-to-door community health worker, the ASHA (Accredited Social Health Activist). The session invites players to play the game on large projection game screens, as the player will perform care and data labour in an unpredictable setting.
Ismael Kherroubi Garcia, Renjie Butalid, Fabio Tollon, Christine R. Kirkpatrick, Ramla Anshur, Connor Wright, Ve Dewey, Tania Duarte, Juliane Schneider and David Figueroa
Coordinator Contact: ismaelkherroubi@gmail.com
“Artificial intelligence” (AI) is often framed as an inevitable force—a future unfolding beyond collective control and concentrated in the hands of a technological elite. But what if AI futures were not something delivered to us, but something we could deliberate, contest, and build together? What if diverse publics could gather to imagine technological futures grounded in real societal and ecological needs, and chart practical pathways toward them? This session invites participants to affirm imagination as a collective practice. Through expert facilitation and participatory world-building, we will move beyond abstract optimism or dystopian critique, and instead co-define actionable, justice-oriented AI futures.
Juliette Zaccour, Sofia Hafner, Alex Edmonds, Luc Rocher, Fadila Sebbane, Anne Devautour, Shelagh McNally and Masako Wakamiya
Coordinator Contact: juliette.zaccour@stcatz.ox.ac.uk
This CRAFT session invites FAccT participants to join AvantAGE, a Montreal-based collective of mature feminist artists, in creating an art installation using old, obsolete cables whilst collectively reflecting on digital ageism and related social justice issues that are exacerbated through technology. Obsolete cables represent (i) aging and ageism, (ii) the loss of physicality in our experiences with computing (with hardware and data being increasingly delocalized), (iii) a reflection on the “retro”, cable-heavy and effortful tech days, and (iv) growing resource consumption and reliance on technology. Weaving represents traditionally feminine and indigenous crafts, often dismissed as art forms. More recently, the emergence of digital crafts and the renewed interest of younger generations in fiber arts has led to conceptualizations of craft as activism and power, labelled craftivism. Like algorithms, weaving ranges from extremely simple and procedural steps to intricate, complex pattern-making. Hence, weaving with cables is an exploration of the connections between physical and digital, old and new, ‘masculine’ objects and ‘feminine’ making. The hands-on cable weaving activity is paired with small-group discussions, threading between collective making and reflecting. The session’s format is inspired by the Stitch’n’Bitch crafting community, centered on creating whilst exchanging critical perspectives. See, Stitch'n'Bitch: Collective reflections on ageism, feminism, and creative resistance through hands-on cable weaving: https://biksil.github.io/craft-2026/
Ulrich Aïvodji, Mina Alfaghih, Meghana Bhange, Elliot Creager, Michelle Lin, Jennifer Mickel, Annie Pullen Sansfaçon, Morgan Klaus Scheuerman, Arjun Subramonian, Jacob Hobbs, Nandhini Swaminathan, Sarah Mathew, Yanan Long, Sabine Weber and Ruchira Ray
Coordinator Contact: ulrich.aivodji@etsmtl.ca
This interactive workshop brings together researchers, practitioners, and community members to explore how LGBTQIA2S+ communities experience and respond to algorithmic harms, and how these responses can inform more inclusive AI governance. Moving beyond top-down approaches, the workshop centers community-driven strategies that enable agency and collective influence over AI systems. Through scenario-based, small-group discussions, participants will identify harms, examine existing coping and resistance practices, and surface unmet needs across domains such as recommendation, moderation, and service access. The workshop aims to foster shared understanding and outline broad, community-centered directions for effective algorithmic collective action, closing the gap between research and LGBTQIA2S+ community needs.
Roya Pakzad and Iranian Women'S Coalition For Internet Freedom Iranian Women'S Coalition For Internet Freedom
Coordinator Contact: rpakzad@taraazresearch.org
Iran's repeated internet shutdowns -- during the 2026 protests and the ongoing Israel/U.S. war -- combined with the role of AI in propaganda, military operations, activism, and daily life, have created an unprecedented crisis at the intersection of AI, censorship, protests, and war. Through 4–5 lightning talks, this session examines what happens when AI-dependent populations lose access during shutdowns; how AI-generated content fuels an information fog that undermines truth and accountability; how AI companies are embedded in military operations with catastrophic consequences; how digital sovereignty rhetoric is being co-opted to justify censorship and control in the AI era; and how LLM safety mechanisms effective in English break down in Farsi. While rooted in Iran, these dynamics are relevant wherever shutdowns, war, protests, and AI intersect, and we invite the FAccT community to engage with and build on this work.
Alicia DeVrio, Inha Cha, Shira Abramovich, Audrey Le Meur, Ali Alkhatib, and David Widder
Coordinator Contact: adevos@andrew.cmu.edu
Despite the work of researchers at FAccT to reform and improve harmful algorithmic systems, such systems have spread into many aspects of digital and public life. In response, affected people have attempted to counter tech power by adopting tactics of resistance and refusal. This craft session highlight what FAccT researchers can gain from understanding resistance to AI. We invite participants to engage with tactics of refusal and resistance toward AI systems, emphasizing that resistance includes a range of situated, strategic responses. Through a collaborative learning exercise, we will provide space for researchers to discuss, reflect on, and share ideas for how we can engage with resistance tactics to support impacted people. We then will reflect together on what resistance can teach us and how our engagement with these acts can support more just and equitable research that more reflexively and meaningfully engages with power.
Vagner Santana, Diogo Cortiz, Beatriz Rocha, Jacqueline Custodio and Henrique Xavier
Coordinator Contact: santana.vagner@gmail.com
Generative Artificial Intelligence (GenAI) and its synthetic outcomes are now a hype(d), omnipresent thing forcing us to continuously ask ourselves: is this real? As GenAI technologies evolve, the hints we still have to grasp on uncanny synthetic content, such as malformed hands in images and homogenized textual content, are growing thin. Hence, in the near future, humans (and machines) will struggle to differentiate real from synthetic, impacting domains such as politics, consumer, and human relationships, potentially rendering GenAI as a Weapon of Math Destruction (O’Neil, 2017), damaging at scale in an opaque way.
Ramaravind Kommiya Mothilal, Faisal Lalani, Dipto Das, Syed Ishtiaque Ahmed, Shion Guha and Sharifa Sultana
Coordinator Contact: ram.mothilal@mail.utoronto.ca
Canada's Directive on Automated Decision-Making requires federal agencies to complete an Algorithmic Impact Assessment (AIA) before deploying automated systems to ensure that decision-making processes are fair and unbiased. While the published AIAs improve public accountability, especially in collecting critical answers to decision-making, they do not explicitly surface the reasoning behind those answers. Most response in an AIA rests on assumptions, things taken for granted and not stated, and these assumptions breed ambiguity about what counts as harm, among others. The goal of this workshop is to bring those assumptions into the open: to name them, examine who they advantage, and ask what it would take to challenge them. Participants will use tools from Informal Logic, the philosophical subfield that studies how everyday arguments actually work, to practice moving from what an AIA says, to what it is taking for granted, to what it is not asking. Participants will also examine what happens when an LLM is asked to do the same thing. The workshop will produce three collective outputs: a publicly available document covering the AIA's hidden assumptions and offering additional questions for advocates, lawyers, and affected communities reading from the outside; a simple analytical tool for examining what any accountability framework takes for granted and who benefits from that; and a shared open research question about how AI accountability frameworks should be redesigned so that reasoning, not just answers, is visible to people outside the institution.
Blair Attard-Frost, Jess Reia and Ana Brandusescu
Coordinator Contact: blair.attardfrost@ualberta.ca
AI systems used for automated decision-making, gender recognition, content generation, and a variety of other applications impose physical, psychological, social, and economic risks upon transgender, nonbinary, and other gender diverse people (collectively referred to in this proposal as “trans”). AI applications already cause trans people to experience harassment, indignity, erasure, social exclusion, coercion, and violence, and the potential for harm has increased against a backdrop of rising anti-trans sentiment in many parts of the world. The impacts that AI systems have on transgender people – as well as the policy interventions and resources that transgender people require to be protected against harmful AI systems – have been largely ignored by AI researchers, practitioners, and policymakers. There is an urgent need to establish a trans-centric AI policy agenda that is made by and for transgender communities.
Building on insights surfaced at a RightsCon 2025 discussion session on transgender issues in AI policy, this workshop will advance a trans community-led agenda for AI policy. This CRAFT session will take the form of a 2-hour policy co-design workshop, during which we will convene a group of 30 researchers, practitioners, policy experts, and community organizers to identify the most pressing impacts that trans people face in AI policy.
Our discussion will be guided by three central questions: Q1 – Impacts: What impacts do AI systems have on trans people? Q2 – Policy gaps: What policy instruments have been created to address those impacts, and what are the gaps and limitations of those policies? Q3 – Recommendations: What actions can researchers, practitioners, policy experts, and community organizers take to address those gaps and limitations?
The workshop will be structured as follows:Diana Mosquera
Coordinator Contact: diana@diversa.studio
This participatory workshop explores the hidden geographies of Artificial Intelligence (AI) by collectively mapping the infrastructures, labor, and territorial dynamics that sustain its development. Challenging dominant narratives that frame AI as abstract and immaterial, the session foregrounds the material conditions and global inequalities embedded in AI systems. Through a structured, hands-on methodology, participants collaboratively map geographies of extraction—including data centers, energy systems, mineral resources, and digital labor—and geographies of resistance, such as community-led initiatives, alternative infrastructures, and practices of data sovereignty. By centering situated knowledge and collective inquiry, the workshop creates a space for participants to critically engage with AI as a socio-technical system shaped by power asymmetries. The resulting maps function both as analytical tools and speculative artifacts, enabling participants to reimagine more just and equitable technological futures. This session contributes to ongoing conversations in critical AI studies by offering a participatory, practice-based approach to understanding and transforming the global political economy of AI.
Soizic Pénicaud, Teresa Barrio Traspaderne, Priya Goswami, Karolina Sztandar-Sztanderska
Coordinator Contact: soizic.penicaud@gmail.com
Governments around the world are increasingly integrating AI and algorithmic decision-making systems into all aspects of social protection and welfare systems, from assessing poverty and vulnerability, verifying applicants’ eligibility, and fraud and corruption detection. A growing body of evidence shows that automated systems can exclude eligible people, intensify surveillance and reproduce discrimination against communities already facing marginalization. Despite these harms, governments and international development actors such as the World Bank frame these tools as improving efficiency and better allocating resources, and continue to promote digital ID systems, digital public infrastructures, and automated decision-making as “best practices” in social welfare. This creates an urgent need to learn from community-led movements and grassroots activists that have raised awareness of these harms and impacts, challenged them politically and legally, and built alternatives. This CRAFT session aims to bring together FAccT researchers with civil society, activists, artists, and community-based practitioners to present case studies of organised resistance to harmful AI in welfare, in Colombia, France, India, and Poland. We will collectively explore:
We will open with a short “grounding” activity to map participant interests and contexts, followed by lightning case studies and Q&A presenting concrete resistance efforts across geographies, actor coalitions, and system types. We will then move into a small-group discussion to compare cases and participant experiences. The discussion will serve as a basis for a concise “resistance tactics and research support” cheat sheet that will be published after the conference.
Natasha D.
Coordinator Contact: natashad.lse@gmail.com
This interactive workshop adopts a speculative world where a large language model (LLM) surveils all digital speech. Attendees play a game to evade censorship, collaboratively designing comedic messages that bypass the moderation system. Participants will learn they can mask their text from censorship by exploiting the context and word order-dependent nature of these models, drawing from “jailbreaking” methods of prompting LLMs with text in the style of poems or wrapping messages in misleading framing. The idea of using visible manipulation is inspired by the security practice of adversarial testing as well as artistic interventions that have illustrated faults in facial recognition technologies.
Humor-based obfuscation makes the harms of LLM-based digital surveillance more tangible and offers a unique approach to algorithm auditing and transparency. Through the collaborative “audit” that participants undertake, the CRAFT session demonstrates how LLMs work, shows that they are not infallible, and affirms that technical expertise is not always required to contest these systems.
Jessica He, Finola Finn, Angel Hsing-Chi Hwang, Donal Khosrowi, Seyun Kim, Morgan Klaus Scheuerman and Runlong Ye
Coordinator Contact: jessicahe@ibm.com
As AI-generated and co-created content proliferates and becomes more difficult to identify, governments and institutions are enacting AI disclosure requirements intended to promote transparency and accountability. However, growing evidence shows that disclosure is not a neutral act: revealing use of AI can lead to stigmatized perceptions of the user and their work, affecting minoritized groups in particular. This CRAFT session immerses participants in the complex benefits, harms, tensions, and power asymmetries that emerge under AI disclosure norms and mandates. Through two participatory activities, attendees will collaboratively analyze how AI disclosure impacts different stakeholders in diverse contexts and speculate on possible futures for transparency that mitigate sociotechnical harms. Participants will leave equipped with concrete methods and design artifacts. These tools will empower them to create community-centered disclosure policies that minimize harm and protect the right to refusal without sacrificing accountability.
Shazeda Ahmed, Joshua Kroll, Brian Merchant, Jacob Metcalf, Tina Park, Andrew Smart and Roel Dobbe
Coordinator Contact: shazeda.i.ahmed@gmail.com
There is a growing assumption in the tech industry, press, and policy circles that “smarter than human” AI is “just around the corner” and that we must prepare for AI systems that are more powerful than humans at core social, scientific, and economic tasks. While progress has been rapid, progress toward what remains ambiguous. We are no closer to understanding what human intelligence fundamentally is, and so what “smarter than human” could mean (and how this creates change in the world) remains ill-defined. Even major figures in the “AGI race” note similar sentiments, with OpenAI CEO Sam Altman stating, “I think [AGI is] not a super useful term ... I think the point of all of this is [AGI] doesn’t really matter and it’s just this continuing exponential of model capability[.]”
What is the cost of pursuing a goal “that doesn’t really matter”? What kind of future is possible if we build AI without AGI? This session invites the FAccT community to begin articulating a vision that decenters AGI/ASI in favor of pro-social technology, and to ask what such an intellectual, research, and policy agenda would look like. This interactive session explores these questions through group discussions and prompting via translational scholarship. The first hour introduces structured provocations interspersed with a “pair-and-share” participant discussions that disrupt the narratives promoted by AGI-enthusiasts and further elaborate upon the theme of “What could a world with AI - not AGI - look like?”:
Silvia Cazacu, Alice Qian, Dora Zhao, Kathleen Pine, Shawn Walker, Hong Shen, Laura Dabbish, Georgia Panagiotidou and Morgan Klaus Scheuerman
Coordinator Contact: silvia.cazacu64@gmail.com
While AI technologies are often framed as ubiquitous and inevitable, their expansion relies on the large-scale extraction of data from diverse global communities. However, the datasets powering foundation models are often treated as found artifacts rather than products of specific power dynamics and human labor. Current practices frequently disregard the structural inequities embedded in data, even as these systems profoundly impact systemically marginalized communities. While frameworks for ethical curation exist for smaller datasets, the massive scale of foundation models has introduced a logic of extraction that prioritizes volume over accountability. This workshop invites researchers, practitioners, and activists to move beyond standard technical hurdles and instead problematize the foundational assumptions of large-scale data work. This workshop builds on a series of ongoing conversations across scholarly communities and continues work initiated at CSCW 2025. Drawing from the CRAFT tradition of transdisciplinary exchange and community action, we will facilitate a collective refactoring of the three core pillars of data curation: composition, focused on interrogating whose lives are extracted and how representation is shaped by hegemonic interests; process, which centers the invisible labor and situated contexts involved in curating and cleaning massive data stores, and release, focused on rethinking the governance, accountability, and potential for refusal in how these models are shared with the world. Our goal is to cultivate a community-led conceptual framework that reimagines more just sociotechnical futures—transforming data curation from a top-down technical requirement into an act of collective responsibility and repair.
Kate Bertash and Rebecca Ackerman
Coordinator Contact: kate@digitaldefensefund.org
"As artificial intelligence is rapidly embedded into everyday workplace software, social justice organizations are facing a quiet crisis of adoption. High-level AI ethics frameworks offer little help to frontline workers struggling with the “Free.99 problem”: the reality that free, highly accessible tech tools often extract a heavy toll in privacy and data harvesting, while safer, more just alternatives are either prohibitively expensive or require immense technical capacity to self-host.
This CRAFT session moves beyond abstract concerns about AI to explore the visceral, daily resource management required to protect movement data and vulnerable communities. ""The AI Just Read Our Discord"" is an interactive, tabletop role-playing simulation designed to immerse participants in the agonizing trade-offs organizations make when navigating vendor risk, shadow IT, and forced AI adoption.
Borrowing mechanics from survival horror tabletop games, participants will break into small groups and collaboratively ""roll up"" a fictional human rights organization using a d6 die. Through a guided character creation process, groups establish fictional but realistic threat models, their core data vulnerabilities, and their starting resources, distributing a limited pool of tokens across three vital stats: Budget, Staff Capacity, and Tech Debt.
Throughout the simulation, the room will collectively face a series of rapid-fire scenario cards drawn from real-world crises observed across 1,000+ organizations, including academic and public institutions, working journalists, researchers, and human rights defenders. Scenarios include sudden funder requirements demanding AI-driven impact metrics, a critical vendor quietly updating their Terms of Service to train models on user data, or exhausted staff adopting an unvetted AI transcription tool to clear a massive backlog.
To survive each 3-minute, high-pressure encounter, groups must debate and spend their limited resources. Do they drain their precious budget for an enterprise license with a zero-retention agreement? Do they burn staff capacity to carefully vet open source alternatives for bias? Or do they accept the ""Free.99"" option and take a critical hit to their community's safety?
By gamifying the friction of tech procurement, this session visualizes the invisible architecture of refusal and makes tangible the hidden tradeoffs we face everyday. It provides a structured, high-energy environment for participants to practice setting boundaries, recognize the resource constraints driving unsafe AI adoption, and explore what true technological agency looks like under pressure.
Anastasia Karagianni
Coordinator Contact: anastasia.karagianni@vub.be
This interactive workshop explores how law, medicine, and AI technology co-produce choreographies of control over the birthing body. Drawing on feminist epistemologies, medical sociology, and dance praxis, it argues that the movement from a social to a medical model of women’s health (Bryers & van Teijlingen 2010) has been succeeded by a new algorithmic model, one that translates embodied experience into codified data rhythms. These “algo-rhythms” regulate whose movements, voices, and sensations are legible within obstetric care. This work uses dance as both metaphor and method (Mulcahy 2021): as metaphor, it frames the clinical encounter as a legal-technological choreography that scripts women’s bodies through risk protocols and predictive analytics; as method, it employs choreographic inquiry to reimagine embodied testimony and resistance to these systems. Building on feminist STS insights (Amir 2023), it shows how algorithmic infrastructures dismiss women’s pain by privileging quantifiable data over lived rhythm and relational movement. Through the lens of dance law, the poster argues that regulation operates not only through statutes and algorithms but through rhythm, the pacing, timing, and sequencing of institutional responses to suffering. Re-tuning these rhythms, it proposes, demands a legal-ethical shift toward embodied listening and algorithmic accountability, so that data systems move with rather than against women’s pain.
Sacha Alanoca, Faye-Marie Vassel, Chijioke Mgbahurike, Archit Lohani and Maroussia Lévesque
Coordinator Contact: sachaa@stanford.edu
The vocabularies that structure AI discourse are not neutral. This interactive workshop treats language as a site of AI resistance and intervention. We argue that contemporary AI discourse is saturated with terminology such as ""existential risk,"" ""foundational models,"" and ""hallucinations"" that has been strategically cultivated and mainstreamed by EA-aligned AI safety communities. Far from neutral, these terms function as epistemic infrastructures: they shape what counts as harm, who holds expertise, and which regulatory interventions appear possible. By contrast, FAccT-oriented work lacks a shared and widely circulating vocabulary, which limits its ability to shape discourse at a comparable scale. This asymmetry leads to the prioritization of speculative, long-term risks while rendering current harms such as surveillance, discrimination, labor exploitation, and environmental costs invisibilized in policy and public debate.
The workshop’s central output is the FAccT AI Solidarity Lexicon, a community-built, living resource that equips researchers, advocates, policymakers, and organizers with vocabulary rooted in lived experience, grounded interventions, and a commitment to fairness, accountability, and transparency. This work will also inform a subsequent white paper and peer-reviewed publication. At a moment when critical AI research faces mounting pressure to cede discursive terrain, this session offers the FAccT community a structured opportunity to build shared language, organize resistance, and reclaim the AI narratives for our collective futures.