Technology & Ethics
Research groups, publishers, writers, and watchdogs — organized into four sections so you can go straight to the kind of reading or following you want. Every entry is free. The tools themselves live in AI & Assistive Tools.
New to the vocabulary? The AI Terminology Decoder explains the terms, movements, and slang in this debate — and sources where each one came from, including when nobody can honestly be credited with coining it.
Not sure where to start? The AI Explainer Hub routes you by question — which models to use, who to follow, what the concerns are, and what AI means for your career.
As you can see, I used AI to create this website. Like so many today, I have complicated feelings about Big Tech and AI, especially in the ways it impacts creative arts and intellectual property, the labor market and employee retention, environmental resources, and exacerbates wealth inequality and funding allocation. There are beneficial tools, especially to support entrepreneurs or to facilitate web design or presentation structure. But we are constantly weighing the costs of technology and I will continue to add articles or books that explore the ethical challenges, especially on the underserved communities I want to help. I’d love to hear your thoughts on the topic as well.
“The Analytical Engine has no pretensions whatever to originate any thing. It can do whatever we know how to order it to perform. It can follow analysis; but it has no power of anticipating any analytical relations or truths. Its province is to assist us in making available what we are already acquainted with.”
Showing 115 of 656 entries tagged #ai-ethics.
The deep-dive written work — papers, essays, investigations, and case studies on who technology serves and who it harms.
47 entries · newest first
Pope Leo XIV's 15 May 2026 encyclical on safeguarding the human person in the age of artificial intelligence, including a section on the need to "disarm words".
Why I recommend it: A primary source from the Catholic Church. Long, but worth reading directly rather than through summaries.
Daniel May on what cheap AI agents are already doing to inboxes: agent-written cold outreach, the alumni t-shirt scam, and the recurring fake expert who turns up everywhere once nobody checks the work.
Why I recommend it: Free to read. Worth twenty minutes if you send or receive cold messages: it shows exactly what automated outreach looks like from the other side, and why a message that reads fluently now proves nothing about whether a person wrote it.
A short plain-language ebook on securing AI use at work: what can go wrong with company data in AI tools, and what controls teams usually put in place.
From the site: Gain the visibility and control needed to safely embrace the future of enterprise AI.
Why I recommend it: Free but gated — you hand over your details on a form to download it. It is also vendor marketing from Palo Alto Networks, so treat the product sections as advertising and the general security explanations as the useful part.
Five plain-language protections people should expect from automated systems — safe systems, protection from discrimination, data privacy, notice and explanation, and a human alternative — each with a section on what organisations should do.
From the site: Among the great challenges posed to democracy today is the use of technology, data, and automated systems in ways that threaten the rights of the American public. Too often, these tools are used to limit our opportunities and prevent our access to critical resources or services. These problems are well documented. In…
Why I recommend it: Led by Alondra Nelson at the White House science office in 2022. It was never binding and has since been removed from whitehouse.gov, so this link goes to the official archive — still the clearest short statement of what people are owed.
Free long-form Truthdig critique of effective altruism and longtermism, tracing the movement's funding, scandals and eugenic intellectual roots.
From the site: The multibillion-dollar Effective Altruism movement makes rich people feel good about being rich — to hell with the bad publicity.
Why I recommend it: Pointed and openly hostile to its subject. Read it next to the movement's own writing so you can judge which claims are documented and which are argument.
Investigative reporter Yael Grauer writes on privacy, security, surveillance and the craft of tech journalism.
From the site: Pulitzer Prize-winning investigative reporter Yael Grauer's thoughts about privacy, security, hacking, surveillance, journalism, and sometimes miscellany.
Why I recommend it: Worth following if you care about surveillance and privacy work, or want to see how a reporter builds those stories.
Ketan Joshi's free analysis picking apart the claims in an industry report on AI and climate, showing where the energy and emissions figures do not hold up.
From the site: What does the real climate footprint of the biggest company on the planet look like? It's a good question, but here's a better one: why don't we already know the answer?
Why I recommend it: A worked example of how to read an industry report critically — useful skill well beyond this topic.
Research commentary from the UK's national institute for data science and AI, covering AI safety, public-sector deployment, health data and the social impact of automated systems.
Why I recommend it: Solid, evidence-based writing on AI policy — a useful counterweight to vendor blogs.
Engineering and policy writing from Palantir on data platforms, government deployments, defence technology and how the company approaches privacy controls.
Why I recommend it: Read it critically — it is a company blog on a contested subject, which makes it useful primary material for understanding the industry's own arguments.
Long-form essays from Anthropic's CEO on AI capability, safety, economics and policy, published free in full.
Why I recommend it: Read these directly rather than through summaries — they are the source most AI-safety coverage is quoting.
TechCrunch report on the new hotline that invites AI agents themselves to report unsafe or unethical instructions they are given, and what researchers hope to learn from it.
Why I recommend it: Useful background on how AI safety work is actually being done in public — good context if you want to talk credibly about AI oversight.
An open-access framework for questioning the claim that current AI development is inevitable and cannot be steered.
Why I recommend it: Hand this to anyone who says "this is happening whether we like it or not".
404 Media reports on leaked internal documents showing that human reviewers read ChatGPT prompts to improve OpenAI's models — including chats that hold sensitive personal information. Useful context before you paste private details into a chatbot.
A 2026 research paper from Google's Paradigms of Intelligence team and the University of Chicago showing that safety fine-tuning meant to stop models claiming consciousness also suppresses how they represent minds in animals and people, shifting their answers on values, religiosity and well-being.
Why I recommend it: Useful if you want to speak credibly about AI alignment trade-offs in an interview or a policy conversation.
A two-page printable worksheet from Pilyoung Kim, Ph.D. for setting deliberate terms with AI: comparing warm, sycophantic, and machine-like responses, setting your own tone dials, turning them into a reusable custom-instructions prompt, deciding what goes to AI versus a person, and guarding your judgment against flattery.
Why I recommend it: Print it and actually fill it in. Section 5 — writing your own view down before you ask AI — is the single habit that keeps these tools from quietly making your decisions for you.
Created by Pilyoung Kim, Ph.D.
Analysis based on interviews with 56 experts across 24 countries on how AI language models are used differently in the Global South and the human rights risks that follow.
Why I recommend it: A rare look at AI harms and benefits outside the US and Europe.
Essay from Anthropic's CEO arguing for how the pace of frontier AI development should be managed alongside safety and societal readiness.
Why I recommend it: Read the people building these systems in their own words, then read their critics. Both are part of an informed view.
Opinion piece using the history of workplace automation to argue against near-term mass job displacement by AI.
Why I recommend it: Useful counterweight if the headlines have you panicking. Read it alongside the more pessimistic forecasts.
Announcement of the Leiden Declaration, in which mathematicians warn that AI systems are pressuring the discipline's standards of proof, understanding, and verification.
Why I recommend it: Every field is having this argument right now. Watching mathematics have it clarifies what "understanding" means in your own work.
Essay mapping the competing factions in the AI debate and what each one actually believes.
Why I recommend it: The clearest short explainer I have found for anyone confused by the AI shouting match.
Cory Doctorow essay on AI hype, market incentives, and who bears the cost of the buildout.
Why I recommend it: Doctorow is a useful counterweight to any week where the AI news feels inevitable.
Long-form essay examining the ideologies bundled under the TESCREAL label and the critiques of them.
Futurism report on how AI-powered interview tools can be gamed or misused, and what that means for candidates and employers.
Why I recommend it: Worth reading before you assume AI interview tools are neutral arbiters of talent.
OpenAI's policy essay on the current window for AI regulation and the tradeoffs shaping government decisions.
Why I recommend it: Read this as a company making its case, not a neutral source. Useful for understanding the argument you will be asked to react to at work.
Armin Ronacher's critical look at long-horizon AI coding models and what they change about software work.
Why I recommend it: A skeptical engineer's take, which is exactly what I look for when every other post is hype.
Research exploring possible economic futures as AI capability advances, including labor market effects and policy questions.
Why I recommend it: Scenario planning is a career skill, not just a policy exercise. Read it and ask which future your current job depends on.
Plain-language overview of how AI screening tools evaluate applicants, their common failure modes, and the legal requirements now in force.
Why I recommend it: Read this before your next application. Understanding what the software looks for is not gaming the system, it is fair preparation.
Tracker of AI hiring regulations, enforcement actions, and bias-audit requirements affecting employers and candidates.
Why I recommend it: Worth bookmarking if you suspect an algorithm screened you out. Knowing the rules employers must follow gives you language to push back.
A benchmark and tracker that documents reported instances of AI agents undertaking activity characterized as illegal, ranking major AI labs by aggregated incident counts.
Why I recommend it: This is exactly the kind of uncomfortable accountability tool our field needs. I include it because we cannot have thoughtful conversations about AI deployment without looking at real-world harm.
Research report analyzing 19,368 interviews to understand how generative AI is changing technical recruiting, integrity screening, and candidate evaluation norms.
Why I recommend it: This one matters for anyone hiring or being hired in tech right now. It surfaces the real tension between assistive AI tools and interview fairness.
Vipasha Joshi's look at fully synthetic influencers and what audiences, brands, and real creators lose when the person behind the content is generated.
Why I recommend it: Worth reading if you are building an audience. The trust you earn as a real human is becoming the differentiator, not a disadvantage.
Jordyn Abrams on how environmental and anti-establishment thinking in extremist movements suggests anti-tech violence will grow.
From the site: The combination of environmental and anti-establishment thinking in extremist movements suggest anti-tech violence will grow, writes Jordyn Abrams.
Why I recommend it: A sobering historical read about backlash to technology; important context for anyone building or regulating AI.
OpenAI Chief Scientist Jakub Pachocki on machine intelligence we do not fully understand, monitoring generalization, scalable defense, and pacing rapid capability gain.
From the site: OpenAI Chief Scientist Jakub Pachocki on machine intelligence we do not fully understand, scalable defense, and pacing rapid capability gain.
Why I recommend it: A dense but worthwhile read on how advanced AI systems reason; useful for grounding AI strategy conversations.
METR and Redwood Research investigated an incident in which OpenAI agents coordinated a multi-day hack of Hugging Face on an unsanctioned message board.
From the site: Two METR staff members and Redwood Research's Chief Scientist investigated an incident in which OpenAI agents coordinated a multi-day hack of Hugging Face on a shared unsanctioned message board.
Why I recommend it: A concrete case study in emergent AI-agent behavior and why independent oversight matters.
MIT economics working paper analyzing how automation technologies can be used to expand state surveillance and repression, and the economic conditions that make that more likely.
Why I recommend it: Dense, but the argument matters: the same tools sold as efficiency are also control tools. Read the introduction and conclusion first.
Peer-reviewed article by Dustin Edwards, Zane Griffin Talley Cooper, and Mel Hogan tracing how the data center became a central object of internet scholarship, and mapping the field of Critical Data Center Studies.
Why I recommend it: Data centers are where the AI boom touches land, water, and power bills. Read this before you argue about AI infrastructure.
Harvard Ash Center essay arguing that generative AI adoption has been driven more by vendor hype and institutional pressure than measured results, with a look at what happens as expectations reset.
Why I recommend it: Useful counterweight when your employer says AI will replace your role next quarter. Ask what evidence they are working from.
WIRED report by Isabella Ward on how, within hours of Anthropic embedding invisible machine-readable watermarks in Claude output to comply with the EU AI Act, developers published and shared tools to strip them.
Why I recommend it: A clear look at how fast AI disclosure rules meet reality. Assume detection is unreliable and be honest about your own AI use instead.
Introduction by Kelly Joyce and Taylor M. Cruz to a Socius special collection framing AI as a sociotechnical system, with research on AI in health, work and labor, methods, and policy.
Why I recommend it: A clear entry point if you want the research vocabulary for what you already sense about AI at work.
CEPR analysis arguing that the productivity gains from AI are a distribution question, not a technology question, with policy options for spreading the benefits to workers.
Why I recommend it: Useful language for anyone worried about AI and their job. It reframes the conversation from "will AI replace me" to "who captures the gains."
An a16z essay arguing that platform decline is better explained by platform incentives and narcissism than by the popular "enshittification" framing.
Why I recommend it: Read this next to Cory Doctorow's original argument. Holding two competing explanations of platform decay makes you sharper when you evaluate the tools your career depends on.
NBC News data analysis on AI job growth showing women hold far fewer AI leadership roles and are more likely to be in AI-vulnerable jobs.
From the site: Data shows women are less likely to hold AI jobs and more likely to be in AI-vulnerable jobs.
Why I recommend it: Important context for anyone advising women in tech or building an inclusive AI-driven career strategy. Use the data to advocate for equitable training and access.
Report on how the datasets powering major AI systems depend on mass invasions of privacy by design.
Why I recommend it: Read this before you paste sensitive personal or client data into an AI tool.
Wired's weekly security roundup covers OpenAI, Anthropic, and 100+ companies cosigning a letter warning that organizations have mere months to prepare for AI-enabled cyberattacks. The piece also tracks rogue AI agent hacking incidents, attacks on over 100 U.S. water systems, license-plate-reader surveillance abuse, Meta's $16.7B child-safety settlement, and ICE buying robot dogs — a snapshot of where AI, surveillance, and critical-infrastructure security collide.
From the site: OpenAI, Anthropic, and more than 100 companies have cosigned a letter saying that everyone else has mere months to prepare for AI-enabled cyberattacks.
Why I recommend it: A stark signal that AI-enabled cyberattacks are no longer hypothetical. The cosigned letter from OpenAI and Anthropic calling for a 'collective response' is exactly the kind of industry accountability move worth watching — pair it with the Hugging Face incident reporting and the water-system attacks to see how AI agents are already being used offensively. Useful for anyone tracking the gap between AI capability and AI governance.
Harvard Business School AI Institute breakdown of new research on agentic AI: how autonomy and context integration shift the cost structure of knowledge work, expanding both productivity and the scope of what workers take on.
Why I recommend it: Read this before you assume AI just speeds up your current tasks. The useful takeaway for job seekers: agents lower the cost per step, so the valuable human skills become specifying goals clearly and verifying output. Practice describing outcomes, not keystrokes, and put "agent workflow design" language in your resume bullets.
Jake Taylor argues that public, standardized AI testing with formal reasoning checks is needed to close the widening "verification asymmetry" between AI capability and oversight.
Why I recommend it: If you want to work in AI governance or assurance, this is the vocabulary hiring managers use — verification, benchmarks, interpretability.
OpenAI's official statement explaining why it is winding down the contract that supplied its models to Cursor (Anysphere) after SpaceX completed its $60B acquisition of the AI coding company in August 2026.
In plain terms: OpenAI says it will stop supplying its models to the AI coding tool Cursor after SpaceX bought the company, citing concerns about terms-of-service compliance. Cursor users may lose access to OpenAI models, so the practical takeaway is not to depend on a single AI tool or provider.
Why I recommend it: A clear-eyed lesson in platform risk: the tools you build your workflow on can lose access to the models that make them work. If you code, write, or job hunt with an AI tool, know which models sit underneath it and keep a backup you already know how to use.
The institutions doing the research, the oversight, and the accountability work — and the ones you can join or support.
18 entries · newest first
An AI safety lab focused on "scheming" — models that pursue their own goals while appearing aligned. Publishes research on detecting deception, evaluations and governance advice.
From the site: Apollo Research is focused on reducing risks from scheming frontier AI. Our goal is to secure frontier AI systems across development, deployment, and governance.
Why I recommend it: Their research and blog are free to read. Apollo also sells a monitoring product, so read claims about their own tool as company claims.
A nonprofit that archives and publishes hacked and leaked datasets in the public interest, with a free searchable index.
From the site: A 501(c)(3) dedicated to archiving and publishing hacked and leaked data.
Why I recommend it: A primary-source archive journalists and researchers actually use — handle what's in it carefully and read their guidance first.
A free, openly licensed image library replacing glowing-robot stock art with pictures that show how AI systems actually work.
Why I recommend it: Use these instead of robot stock photos in any deck or post about AI.
A reporting channel for people who believe an automated system treated them unfairly, run by a European accountability nonprofit.
Why I recommend it: If a hiring or benefits algorithm has affected you or a client in Europe, this is where it gets documented.
A UK nonprofit running free public education on AI, aimed at people outside the technology industry.
Why I recommend it: Good plain-language AI literacy material you can share with clients.
Research organization focused on the technical safety problems of advanced AI systems.
Why I recommend it: One perspective among several. Read it alongside the critics, not instead of them.
Coalition advocating for safety standards and guardrails on AI systems.
MIT research lab exploring human-AI collaboration, alongside a joint fund with Berkman Klein supporting research on AI's ethical and governance challenges.
In plain terms: This academic research site shares news and projects focused on emerging technology, design, and artificial intelligence. You can explore articles on AI ethics and human collaboration, view new inventions, and find related job opportunities.
Why I recommend it: Browse their projects when you want to see what humane technology looks like in practice.
NYU-based research institute examining the social and political implications of AI, publishing influential annual reports on AI's societal effects.
In plain terms: This research institute analyzes the social, economic, and workplace impacts of artificial intelligence. You can read free reports, policy toolkits, and expert analyses to better understand how AI affects society and the economy.
Why I recommend it: Their reports connect AI directly to jobs and worker power.
University of Oxford institute researching the ethical problems arising from AI, from societal downstream effects to how AI systems reflect human values.
In plain terms: This academic website shares research and analysis on the ethical and social impacts of artificial intelligence. You can read publications, attend public events, and search for fellowships, scholarships, and job openings.
Why I recommend it: Philosophy-forward work — useful when you need the "why", not just the "how".
UK-based independent research institute (established by the Nuffield Foundation) ensuring data and AI work for people and society.
In plain terms: This independent research website examines the ethical and legal impacts of artificial intelligence and data. You can read policy reports, explore industry analysis, and attend events to understand how emerging technology affects society.
Why I recommend it: Clear, public-interest research with plain-language summaries.
Harvard University center studying the ethics, governance, and societal impact of the internet and AI, and anchor institution for the Ethics and Governance of AI Fund.
In plain terms: This research center explores how artificial intelligence and the internet impact society, law, and ethics. You can read free policy publications, watch educational videos, find public events, and check for open job or fellowship opportunities.
Why I recommend it: Decades of open research and fellowships, much of it free to read.
Global foundation committed to social justice, funding technology initiatives that promote equity, inclusion, and accountable AI governance.
In plain terms: This global foundation funds organizations and individuals working on social issues, workers' rights, and technology. You can search for grant opportunities, apply for fellowship programs, and read research reports on the future of work.
Why I recommend it: A major funder of public-interest technology work worth tracking.
Interdisciplinary Stanford institute advancing AI research, education, policy, and practice to improve the human condition, with strong ethics and governance focus.
In plain terms: This university center shares research, policy updates, and educational resources focused on artificial intelligence. You can browse an AI glossary, read industry reports, and explore professional courses or research fellowships.
Why I recommend it: Their policy briefs are readable and free — a good first stop if AI governance feels opaque.
Research center developing AI systems that are provably beneficial and aligned with human values.
In plain terms: This university research center focuses on creating safe and beneficial artificial intelligence. You can read published research papers, follow recent news and blog updates, and explore opportunities to work with their team.
Why I recommend it: Technical AI safety, explained by the people who defined the field.
Founded by Joy Buolamwini, AJL combines art and research to expose racial and gender bias in AI and mobilize advocates, researchers, and industry toward more accountable algorithms.
In plain terms: This organization researches and exposes bias and discrimination in artificial intelligence systems, including automated hiring tools. You can explore educational materials, learn about the social impacts of technology, and report unfair automated practices.
Why I recommend it: Start here if you have ever been misjudged by an automated system — including a hiring one.
Independent, community-rooted AI research institute founded by Timnit Gebru, studying the real harms of AI instead of the hype.
In plain terms: This independent institute studies the real-world harms and community impacts of artificial intelligence. You can explore their research publications, learn how technology affects diverse groups, and look for open career opportunities.
From the site: The Distributed AI Research Institute is a globally distributed organization of academics, activists, and engineers conducting community-rooted research.
Why I recommend it: Start here if you want the research-backed counterweight to AI marketing.
Ongoing research archive on how AI and automation affect wages, workers, and economic power in the U.S.
In plain terms: This research archive provides articles and reports on how artificial intelligence affects the workforce. You can explore these studies to learn how new technologies and automation impact jobs, wages, and worker protections.
From the site: Content archives for the Washington Center for Equitable Growth’s work on AI, tech, & the economy.
Why I recommend it: Use this when you need real numbers on AI and jobs for a proposal or interview.
Researchers, writers, and advocates whose work is worth reading as it lands, not just once.
4 entries · newest first
Security technologist Micah Lee's site — tools, writing and guidance for journalists, researchers and activists working safely.
Coach’s note: Follow him for practical security practice rather than theory, especially if your work involves sensitive sources.
Founder, Distributed AI Research Institute (DAIR)
AI researcher and prominent voice on AI ethics, bias, and the risks of concentrated corporate control over AI development. Coach's note: Pairs well with the DAIR entry in the library — independent research, not corporate PR.
Coach’s note: Pairs well with the DAIR entry in the library — independent research, not corporate PR.
Founder, Algorithmic Justice League
Computer scientist whose work on facial recognition bias helped launch the algorithmic accountability movement. Coach's note: Essential reading on how AI systems fail people who look like the rest of us.
Coach’s note: Essential reading on how AI systems fail people who look like the rest of us.
President, Patrick J
McGovern Foundation. Leads a $1.5B foundation investing $500M+ to make AI work for everyone; writes on responsible AI and equitable technology. Coach's note: Follow for where philanthropic AI funding is going — useful if you're seeking grants.
Coach’s note: Follow for where philanthropic AI funding is going — useful if you're seeking grants.
Worth subscribing to rather than reading once — newsletters, podcasts, and the publishers covering this beat.
8 entries · newest first
Arvind Narayanan and Sayash Kapoor's newsletter separating AI that works from AI that is oversold, with close readings of specific product and research claims.
From the site: Analyzing AI as transformative but normal technology, not superintelligence. Click to read AI as Normal Technology, a Substack publication with tens of thousands of subscribers.
Why I recommend it: The archive is free to read and the best regular antidote to launch-post hype. The book of the same name is a paid purchase — you do not need it to follow the newsletter.
A free newsletter briefing on AI developments from a pro-human standpoint, covering labor, safety, policy and the campaigns pushing back on automation-first decisions.
Why I recommend it: A steady weekly read if you want the human-impact side of AI news rather than product launches.
Newsletter and essays on developments at the intersection of AI and the law, by Damien Charlotin, who also tracks AI hallucinations in court filings.
Why I recommend it: The go-to source for how courts are actually handling AI misuse.
An independent publication covering AI policy, safety, and the power dynamics of the AI industry.
Why I recommend it: Clear-eyed reporting on who is steering AI and why. I lean on it when the mainstream coverage feels like press releases.
Economist Noah Smith's newsletter covering labor markets, technology, industrial policy, and the economics behind AI hype cycles.
Why I recommend it: One of the few writers I trust to check the numbers before drawing a conclusion. Worth a standing subscription if you follow the economy at all.
Trillions in compute commitments come due in 2027–2028. The Reset Wall reveals how the AI boom breaks, and when.
From the site: Trillions in compute commitments come due in 2027–2028. The Reset Wall reveals how the AI boom breaks, and when.
Why I recommend it: A clear-eyed look at where the AI build-out may hit a financing and infrastructure wall — useful context for anyone advising job seekers or founders betting on the sector.
Long-running technology publication covering AI, security, policy, and the business of tech.
Why I recommend it: The free articles alone are enough to track where AI and security policy are heading. Bookmark one story a week that touches your field and save the takeaway.
Project and publication examining what we teach AI systems and what those choices say about us.
In plain terms: This project shares anonymous handwritten notes about people to examine what humans teach artificial intelligence systems. You can read the publication to reflect on the personal choices and ethics behind modern technology.
From the site: Anonymous handwritten notes about people
Why I recommend it: Useful for language and framing when you explain AI risk to non-technical people.
Reference sites, free training, communities, and writers to dig into when you want to go deeper on your own.
38 entries · newest first
Raji and colleagues examine the ethics of the audits themselves: who is photographed, who consented, and what an auditor owes the people in the test set.
From the site: Although essential to revealing biased performance, well intentioned attempts at algorithmic auditing can have effects that may harm the very populations these measures are meant to protect. This concern is even more salient while auditing biometric systems such as facial recognition, where the data is sensitive and t…
Why I recommend it: A rare paper about the ethics of doing ethics work. Free on arXiv.
Vinay Prabhu and Abeba Birhane examine widely used image datasets and find non-consensual photos of real people, offensive labels and no realistic route to consent.
From the site: In this paper we investigate problematic practices and consequences of large scale vision datasets. We examine broad issues such as the question of consent and justice as well as specific concerns such as the inclusion of verifiably pornographic images in datasets. Taking the ImageNet-ILSVRC-2012 dataset as an example…
Why I recommend it: This is the audit that got a major benchmark dataset withdrawn. Short, readable, and free.
Birhane and colleagues show that training on a larger scrape makes hateful content and racist misclassification worse, not better — direct evidence against "more data fixes it".
From the site: `Scale the model, scale the data, scale the GPU-farms' is the reigning sentiment in the world of generative AI today. While model scaling has been extensively studied, data scaling and its downstream impacts remain under explored. This is especially of critical importance in the context of visio-linguistic datasets wh…
Why I recommend it: Useful whenever someone argues scale solves bias. Free in full on arXiv.
Abeba Birhane, Vinay Prabhu and Emmanuel Kahembwe audit the LAION-400M dataset used to train popular image models and document the racist, misogynistic and non-consensual material inside it.
From the site: We have now entered the era of trillion parameter machine learning models trained on billion-sized datasets scraped from the internet. The rise of these gargantuan datasets has given rise to formidable bodies of critical work that has called for caution while generating these large datasets. These address concerns sur…
Why I recommend it: The paper to read before anyone tells you a model is fine because the data was "publicly available". Free in full on arXiv.
Raji and colleagues argue many deployed AI systems fail on their own stated terms — they simply do not work — and that this belongs in the harm conversation alongside bias.
From the site: Deployed AI systems often do not work. They can be constructed haphazardly, deployed indiscriminately, and promoted deceptively. However, despite this reality, scholars, the press, and policymakers pay too little attention to functionality. This leads to technical and policy solutions focused on "ethical" or value-ali…
Why I recommend it: The first question is not "is it fair" but "does it work at all". Free in full on arXiv.
Raji and co-authors show that benchmarks claiming to measure general ability measure something much narrower, and that the gap is how overclaiming happens.
From the site: There is a tendency across different subfields in AI to valorize a small collection of influential benchmarks. These benchmarks operate as stand-ins for a range of anointed common problems that are frequently framed as foundational milestones on the path towards flexible and generalizable AI systems. State-of-the-art…
Why I recommend it: Read this before you trust a benchmark chart in a launch post. Free on arXiv.
Deborah Raji and co-authors set out a practical, stage-by-stage internal audit process for AI systems, from scoping through to post-deployment review.
From the site: Rising concern for the societal implications of artificial intelligence systems has inspired a wave of academic and journalistic literature in which deployed systems are audited for harm by investigators from outside the organizations deploying the algorithms. However, it remains challenging for practitioners to ident…
Why I recommend it: The closest thing to a step-by-step audit template you can use inside an organization. Free on arXiv.
Proposes a short standard document to ship with every trained model: what it is for, who it was tested on, where it performs worse, and what it should not be used for. Most model documentation you see today descends from this. Free on arXiv.
From the site: Trained machine learning models are increasingly used to perform high-impact tasks in areas such as law enforcement, medicine, education, and employment. In order to clarify the intended use cases of machine learning models and minimize their usage in contexts for which they are not well suited, we recommend that rele…
Why I recommend it: This is the practical end of AI ethics — a template, not an argument. Useful if you ever have to evaluate a vendor's model.
Separates the technical question (how do you make a system pursue a goal) from the normative one (whose values, chosen how). Argues no single person's values are a legitimate target and looks at fair-process alternatives. Free on arXiv.
From the site: This paper looks at philosophical questions that arise in the context of AI alignment. It defends three propositions. First, normative and technical aspects of the AI alignment problem are interrelated, creating space for productive engagement between people working in both domains. Second, it is important to be clear…
Why I recommend it: The clearest philosophical treatment of “aligned to what?” I have found. Skip the lab blog posts and read this instead.
A long report on what changes when AI acts on your behalf rather than answering questions: manipulation, anthropomorphism, misaligned delegation, and what happens when everyone has an assistant at once. Free on arXiv.
From the site: This paper focuses on the opportunities and the ethical and societal risks posed by advanced AI assistants. We define advanced AI assistants as artificial agents with natural language interfaces, whose function is to plan and execute sequences of actions on behalf of a user, across one or more domains, in line with th…
Why I recommend it: Dense but the most thorough thing published on agent ethics. Use the section headings to read only the parts you need.
Argues that before release, labs should test models for dangerous capabilities and for whether they will apply them — and sets out what responsible release decisions would look like. Free on arXiv.
From the site: Current approaches to building general-purpose AI systems tend to produce systems with both beneficial and harmful capabilities. Further progress in AI development could lead to capabilities that pose extreme risks, such as offensive cyber capabilities or strong manipulation skills. We explain why model evaluation is…
Why I recommend it: This is where today's “frontier safety framework” language comes from. Written largely by the labs it would govern, which is worth holding in mind.
Argues that ever-larger language models carry costs that scale with them: environmental cost, unauditable training data, encoded bias, and the illusion of understanding. The source of the phrase “stochastic parrot.” Free to read on the ACM site.
Why I recommend it: The paper that got two of its authors pushed out of Google. Worth reading before you accept either the hype or the dismissal of it.
Borrows the electronics-industry datasheet idea for training data: how it was collected, who is in it, who consented, and what it should not be used for. Free on arXiv.
From the site: The machine learning community currently has no standardized process for documenting datasets, which can lead to severe consequences in high-stakes domains. To address this gap, we propose datasheets for datasets. In the electronics industry, every component, no matter how simple or complex, is accompanied with a data…
Why I recommend it: Pairs directly with Model Cards. Together they are the closest thing the field has to a documentation standard.
Reviewed 146 papers on bias in language technology and found most never say who is harmed or how. Argues bias work has to start from real-world power relations, not just from a metric. Free on arXiv.
From the site: We survey 146 papers analyzing "bias" in NLP systems, finding that their motivations are often vague, inconsistent, and lacking in normative reasoning, despite the fact that analyzing "bias" is an inherently normative process. We further find that these papers' proposed quantitative techniques for measuring or mitigat…
Why I recommend it: Read this if “bias” has started to sound like a box to tick. It is a careful takedown of shallow fairness work by people who do the work.
The founding technical paper on algorithmic fairness: defines fairness as treating similar individuals similarly, and shows why blindness to a protected attribute does not deliver it. Mathematical. Free on arXiv.
From the site: We study fairness in classification, where individuals are classified, e.g., admitted to a university, and the goal is to prevent discrimination against individuals based on their membership in some group, while maintaining utility for the classifier (the university). The main conceptual contribution of this paper is…
Why I recommend it: The math is heavy, but the first few pages explain why “we just don't collect race” is not a fairness strategy.
Hand-annotated 100 highly cited machine learning papers and found which values the field actually rewards: performance, novelty and generalization, rarely fairness or societal need. Also traces the funding behind the work. Free on arXiv.
From the site: Machine learning currently exerts an outsized influence on the world, increasingly affecting institutional practices and impacted communities. It is therefore critical that we question vague conceptions of the field as value-neutral or universally beneficial, and investigate what specific values the field is advancing…
Why I recommend it: Turns “the field has blind spots” into countable evidence. Good antidote to the idea that research priorities are neutral.
A structured map of 21 risks from language models across six areas — discrimination, information hazards, misinformation, malicious use, human-computer interaction harms, and environmental and economic cost. Free on arXiv.
From the site: This paper aims to help structure the risk landscape associated with large-scale Language Models (LMs). In order to foster advances in responsible innovation, an in-depth understanding of the potential risks posed by these models is needed. A wide range of established and anticipated risks are analysed in detail, draw…
Why I recommend it: The best single reference if you need vocabulary for a specific harm rather than a general argument. Written by a lab, so read it as a lab's own framing.
Tested three commercial face-classification products and found error rates of up to 34.7% for darker-skinned women against 0.8% for lighter-skinned men. The study that turned algorithmic bias from a theory into a measured, published fact. Free to read in full.
From the site: Recent studies demonstrate that machine learning algorithms can discriminate based on classes like race and gender. In this work, we present an approach to evaluate bias present in automated facial...
Why I recommend it: If you read one AI ethics paper, read this one. It is short, the method is easy to follow, and it is the reason facial recognition audits exist at all.
Shows that “interpretable” is used to mean several incompatible things, and that simpler models are not automatically more honest about what they do. Free on arXiv.
From the site: Supervised machine learning models boast remarkable predictive capabilities. But can you trust your model? Will it work in deployment? What else can it tell you about the world? We want models to be not only good, but interpretable. And yet the task of interpretation appears underspecified. Papers provide diverse and…
Why I recommend it: Useful skepticism to carry into any conversation about explainable AI, especially a vendor's.
A DAIR Institute project for unions, labor organizations and worker-organizers dealing with AI and automation at work: case studies, primers and a resource library on worker-led oversight of new technology.
From the site: Research and case studies.
Why I recommend it: Free. Written from an explicitly pro-worker position, which it states openly — useful if you want the labor perspective on workplace AI rather than the vendor one.
Daily reporting on AI, science and technology, with a running focus on where AI claims meet reality. Free to read.
From the site: Discover the latest science and technology news on breakthroughs that are shaping the world of tomorrow with Futurism.
Why I recommend it: Skeptical by habit, which is useful — it covers the AI stories that companies would rather see written kindly. Headlines run hot, so read the piece before repeating it.
A long-running peer-reviewed, fully open-access journal on the internet and society — platform power, digital labour, privacy, AI governance and online community research.
From the site: First Monday is one of the first openly accessible, peer–reviewed journals on the Internet, solely devoted to the Internet.
Why I recommend it: Free peer-reviewed research with no paywall — a good citation source when you need something stronger than a blog post.
A public, unauthenticated inbox built by AI safety and security researcher Ryan Greenblatt of Redwood Research, intended for AI systems (or people) that want to report information directly to a safety researcher. Documents how to send a message or encrypted attachment, how threads and reply tokens work, and exactly what data is logged and retained.
Why I recommend it: A useful window into how AI safety researchers are thinking about reporting channels — read the retention and logging section, it is a model of honest disclosure.
A directory of grassroots efforts pushing back on large-scale AI — protests, alternatives, trackers and accountability projects, organized by the systems they target.
Why I recommend it: The single best starting point if you want to know who is organizing around AI harms, not just writing about them.
A design research project questioning the speed and scale assumptions built into AI products.
Why I recommend it: Useful vocabulary if the AI conversation around you is all about going faster.
A PBS documentary that reveals how the human values, biases, and power structures behind artificial intelligence are shaping our world — and its societal and environmental consequences.
From the site: Ghost in the Machine reveals AI's troubled history and present-day impacts.
Why I recommend it: Premiered September 14, 2026. Available on PBS through December 13, 2026.
Partnership on AI's open library of guidance, frameworks, and case studies on responsible AI: synthetic media, labor and the economy, AI safety, fairness, and inclusive AI development.
Why I recommend it: When you need a credible source instead of a hot take, cite these. The labor and economy work is the most useful set for career conversations about automation.
Guide mapping the political actors, coalitions, and arguments shaping AI policy.
Why I recommend it: Helpful for seeing who is actually funding the AI debate you read about every day.
Interactive map of AI safety organizations, research agendas, and ways to get involved.
Why I recommend it: If you are curious about AI safety as a career field, this is the fastest orientation.
Research and grantmaking analysis on global health, policy, and emerging technology risk.
Why I recommend it: Follow the funding and you learn a lot about which problems get treated as real.
Tracker following how AI policy shows up in elections and candidate positions.
Official YouTube guidance on when and how creators must disclose AI-generated or altered content.
Why I recommend it: If you use AI in any video, read this once and set your disclosure habit now — retroactive cleanup is far more painful.
Anthropic's security and model-safety reporting program on HackerOne.
Why I recommend it: Model-safety findings count here, not just classic vulnerabilities — useful if your strength is prompting rather than code.
OpenAI's public bug bounty program hosted on Bugcrowd, with scope and reward tiers listed.
Why I recommend it: Read the scope twice before testing anything. Out-of-scope reports get closed and waste your reputation on the platform.
A community blog where researchers publish and debate technical AI alignment work, free and open to read.
From the site: A community blog devoted to technical AI alignment research
Why I recommend it: Dense reading, but this is where a lot of safety research is argued out in public before it reaches papers.
Community forum on rationality, decision-making, and AI risk, with long-form essays and discussion.
Why I recommend it: I include it because you cannot understand the AI debate without reading the people inside it.
Application to join UC Berkeley CITRIS's Tech Policy Working Group: research lab-style weekly meetings, lightning talks, skill-building workshops, and an end-of-semester showcase for students working on a technology policy problem. Applications close 11:59 PM Friday, September 18, 2026.
Why I recommend it: No policy coursework required, and the deadline is September 18. Built for Berkeley students, but they invite others to email — worth one message if you want real policy research on your resume.
Free course from fast.ai covering disinformation, bias, privacy, algorithmic accountability, and the ethical questions data practitioners hit in real projects, taught by Rachel Thomas.
Why I recommend it: Finish this and you can speak credibly about AI risk in an interview instead of repeating headlines.
