Humane Intelligence

Our UNGA 81 side-event recap is now available, including the full event recording! The event brought together people from across disciplines for insightful conversations about AI safety and our collective future. Learn more: humane-intelligence.org/unga-81-side...

Who gets to decide what counts as harmful in AI? In our latest blog post, volunteer Lindsay Langenhoven explores how open source AI red teaming can help shift that balance of power, enabling local communities to define and test AI risks on their own terms. humane-intelligence.org/post/democra...

Our methodology paper on ontology-based contextual AI evaluations is now available, developed to address 3 challenges: - Lack of scientific rigor arising from unclear testing coverage - Balancing human expertise and automations - Reproducibility of the testing environment arxiv.org/abs/2610.005...

What does it really take to rigorously evaluate an AI system? Our Executive Director, Mala Kumar, joined The Human Protocol podcast to talk about adversarial testing, how evaluators look for where models fail, and the growing role of human judgment in AI evaluation. www.youtube.com/watch?v=_LVk...

Free access to our LinkedIn Learning course continues through September 30, 2026. Thanks to our friends at LinkedIn, AI Evaluations for Everyone: How Non-Engineers Can Build Better AI Systems is currently available free to all learners. Please share it with with your network: lnkd.in/ghgKxn7c

Have you signed up for our $10,000 Bias Bounty Mapping Equity Challenge with Zindi? The challenge asks an important question: Are communities most vulnerable to climate risk also among the least well-mapped? Registration is open to all through 10/31/26. Learn more and register: lnkd.in/gprZnXZb

Want to work with Humane Intelligence and help us continue to grow? See our latest funding request deck, which lays out what we’re asking for and why. If you value our work we’d love for you to take a look. You can also help by sharing it or sending it to funder contacts. lnkd.in/g58_iP6P

Our Bias Bounty Mapping Equity Challenge launches on 6/28! The challenge asks whether communities most vulnerable to climate risk are also among the least well-mapped. Learn more and sign up: zindi.world/competitions...

Are the communities most vulnerable to climate risk also the least well-mapped? Our Bias Bounty Mapping Equity Challenge on Zindi invites participants to explore this question. Registration is open! zindi.africa/competitions...

We're launching a new LinkedIn Learning course titled "AI Evaluations for Everyone: How Non-Engineers Can Build Better AI Systems". The course will be available 07.01 and thanks to our friends at LinkedIn, will be available for free until 09.30! humane-intelligence.org/post/linkedi...

The future of AI depends on data but who decides how that data can be collected, shared, and used? Read more in our latest blogpost: humane-intelligence.org/post/the-glo...

What happens when two AI models invent the same wrong answer? Our latest blog post explores “folie à deux in the machine,” a case of cross-model hallucination overlap that may reveal gaps in how we evaluate and govern AI systems. Read more: humane-intelligence.org/post/folie-a...

What happens when AI evaluation asks not just “Is this accurate?” but “Accurate for whom?” Our Bias Bounty on Zindi will explore geospatial data gaps, climate risk, and a new Bias Scoring API designed to measure model disparities across communities. humane-intelligence.org/post/bias-bo...

Last year we shared that with support from Google.org, we would be releasing an improved version of our AI red teaming app under an open source software license. Since then, a lot of work has gone into moving the project forward. Read more: humane-intelligence.org/post/updates...

Where does mental health fit in how we design and evaluate AI systems? As AI becomes more embedded in daily life and work, its psychological impact—on both users and the people behind these systems—is becoming harder to ignore. Read our recent blog post: humane-intelligence.org/post/mental-...

We’re excited to welcome Amy Sample Ward, Amy S., and Abby V. to the Humane Intelligence Board of Directors, and Dr. Lauren Damme, PhD and Dr. Diego Garcia-Olano to our Advisory Group. Read our latest blog post to get to know them: humane-intelligence.org/post/our-new...

We recently facilitated an AI evaluation workshop at the Tribal Innovation Summit. Adversarial stress testing participants exposed vulnerabilities in AI models and assessed them against Indigenous data governance principles. See the report under "Research." humane-intelligence.org/insights/

A recent report from National Institute of Standards and Technology explores we assess real-world risks of AI systems combining model testing, red teaming and field testing to understand how AI systems behave in use. We're proud to have contributed to this work. nvlpubs.nist.gov/nistpubs/ai/...

On April 14, we’re joining a conversation at the sidelines of the 59th Session on the Commission on Population and Development. Mala Kumar and Annie Brown will contribute to a discussion on ontologies, generative AI, and technology-facilitated gender-based violence. Register: lnkd.in/gpujupu9

We are pleased to share the Indigenized Adversarial Stress Testing Framework, developed by Emergence Circle, with the support of @humaneintelligence.bsky.social! This project has tested AI systems against Indigenous values, community priorities, and potential harms before deployment or adoption.

How do you know if an AI system performs well in a real-world use case? Evaluating AI isn’t just about accuracy or benchmarks but understanding context, scope, and coverage—what’s being tested, what’s missing, and how those gaps shape our conclusions. humane-intelligence.org/post/knowled...

This Women’s History Month, we wanted to share a more personal look at the women of HI: their inspirations, unexpected talents, and the experiences that shaped their paths. A reminder that building better technology ultimately starts with understanding people. humane-intelligence.org/post/the-wom...

Last year, we partnered with the CoNA Lab at Virginia State University and Valence AI to launch the Global Accessibility Bias Bounty. In our blogpost, we discuss why this work matters and how to evaluate and build in a way that's inclusive. humane-intelligence.org/post/reflect...

Humane Intelligence is growing, and we are excited to share updates to our Board of Directors and Advisory Group. We welcome Admas Kanyagia and Benjamin Kinsella to our Board, and Michael Zargham and Kaitlin Thaney to our Advisory Group. humane-intelligence.org/post/our-new...

Join Humane Intelligence and Tattle Civic Technologies at the India AI Impact Summit. Our main event, Evaluations and Open Source Software for AI for Social Good at Scale will be on February 20, 2026, 1:30–2:25 PM IST. This session is free, sign-up required. impact.indiaai.gov.in/registration

We’re still welcoming submissions from interested organizations for our paid opportunity related to the AI red teaming app. Applications are reviewed on a rolling basis, and the deadline to apply is February 6, 2026. Details and the submission form are below. Thanks for helping us spread the word.

In a new guest blog authored by Humane Intelligence volunteers Károly Boczka and Nkechika Ibe, we explore how large language models struggle with bias, distortion and hallucination in low-resource languages and why this is becoming a governance issue. humane-intelligence.org/post/ai-gove...

As the SDGs near 2030, what comes after? In this blog post, we explore how generative AI and ontologies could provide a more connected, flexible lens on human development beyond today’s SDG taxonomy. humane-intelligence.org/post/ontolog...

As AI systems scale, the physical infrastructure behind them matters more than ever. In a guest blog post, we explore how the concentration of data centers and compute power is reshaping geopolitics, governance, and AI evaluation practices. humane-intelligence.org/post/changin...

We’re still welcoming submissions from interested organizations for our paid opportunity related to the AI red teaming app. Applications are reviewed on a rolling basis, and the deadline to apply is February 6, 2026. Details and the submission form are below. Thanks for helping us spread the word.

Humane Intelligence has launched an expression of interest to hire an engineering firm for the backend Python development of our OSS AI red teaming app (paid opportunity. Full description: docs.google.com/document/d/1... Submit your interest here: docs.google.com/forms/d/e/1F... Thank you!

As we close out 2025, we’re reflecting on a pivotal year for Humane Intelligence. From major funding milestones to new strategy, expanded bias bounties, red teaming work, and global partnerships, it was a year of growth. Read our full year-in-review: humane-intelligence.org/post/humane-...

We're joining forces with @humaneintelligence.bsky.social to move their bias bounty program onto Zindi, a data science platform with users in more than 185 countries, with @heisingsimonsfdn.bsky.social's support. Read the announcement here: humane-intelligence.org/post/announc...

Great news! Humane Intelligence has received funding from Google.org to accelerate the release of an open source AI red teaming application. This support will help expand access to participatory AI evaluations worldwide. Learn more: humane-intelligence.org/post/ai-red-...

We’re partnering with Radiant Earth and Zindi, with support from the Heising-Simons Foundation, to bring our bias bounty program to Zindi’s global data science platform! Learn more: humane-intelligence.org/post/announc...

Following our 2024 TFGBV red-teaming work with UNESCO, our Playbook on red teaming AI for social good is now available in French and Spanish! Sharing as 16 Days of Activism begins. EN: unesdoc.unesco.org/ark:/48223/p... FR: unesdoc.unesco.org/ark:/48223/p... SP: unesdoc.unesco.org/ark:/48223/p...

We are seeking volunteers to help redesign the user interface of our TFGBV taxonomy and ontology website. We welcome contributions at any level, whether that is proposing new workflows, creating wireframes or prototypes, or building a front end. Learn more: humane-intelligence.org/get-involved...

AI in public health remains one of the most overlooked areas in the current wave of AI investment. In her op ed for Security Brief TechDay, Mala Kumar, explains how Humane Intelligence is working to address it through our new AI in Public Health Working Group. securitybrief.com.au/story/why-ai...

Today is the last day to submit a response to our bias bounty on accessibility! Please find here all relevant info: humane-intelligence.org/programs-ser...

According to a recent CNBC article, people with ADHD, autism, and dyslexia report that AI assistants are helping them thrive at work. That’s exactly what our Bias Bounty 4 is all about! - Submissions close this Friday! - Join here: lnkd.in/gfdtHTij - Read the article: lnkd.in/erq96y-8

Announcing a new working group on AI in Public Health! Humane Intelligence is collaborating with Rumi Chunara, who leads New York University’s Center for Health Data Science, to launch a new working group focused on exploring the role of AI in public health. humane-intelligence.org/post/join-ou...

Deadline extended! Submissions for the Accessibility Bias Bounty Challenge are now due November 14 at 11:59 PM ET. With Design and Data Science tracks and a $6,000 prize pool, participants are invited to build AI tools that prioritize impacted communities. humane-intelligence.org/programs-ser...

There’s still time to join Bias Bounty 4: Accessibility in Digital Conferencing Facilities! - Design and Data Science tracks - $6,000 prize pool - Contribute to a growing community of practice solving technical bias challenges Submissions close 11/7. humane-intelligence.org/programs-ser...

Annie Brown, founder of Reliabl and Humane Intelligence’s Bias Bounty Data Scientist, joins Brent O. Phillips on the Humanitarian AI Today Voices flashpod to discuss our Accessibility Bias Bounty Challenge Set, open now through November 7, 2025. Tune in here: soundcloud.com/humanitarian...

In a piece for the Columbia Climate School, Marco Tedesco explores this question through the lens of our recent UNGA80 side event, “Honest Discussions at the Intersection of AI and the SDGs,” co-hosted by Humane Intelligence, Technology Salon and Compiler. lnkd.in/g9vGveVd

As we consider the current and future impact of AI, we’re proud to join forces with @partnershipai.bsky.social, a coalition bringing diverse voices together to ensure developments in AI benefit people and society. Learn more about our partnership: partnershiponai.org/partnership-...

We’re proud to celebrate Dr. Julie Hollek, Humane Intelligence’s Head of Engineering, for being named to the CTO Craft 100, recognizing 100 technology leaders shaping the future of the industry! Congratulations, Julie, on this well-deserved recognition! See the full list : lnkd.in/gnd3wQ6p

Today we are launching the Accessibility in Digital Conferencing Facilities Bias Bounty Challenge in partnership with the CoNA Lab at Virginia State University and Valence AI. - Design and Data Science tracks - $6,000 prize pool Submissions close 11/7 humane-intelligence.org/programs-ser...

Want to learn more about our work, get involved in our latest Bias Bounty, or explore volunteer opportunities with Humane Intelligence? Join one of our upcoming Open Community Calls! 📅 Early Edition: 9:00–10:00 AM ET lnkd.in/gHweJHnE 📅 Late Edition: 12:00–1:00 PM ET lnkd.in/gEvKmvYX

At #UNGA80, Humane Intelligence, Technology Salon, and Compiler co-hosted a side-event on Honest Discussions at the Intersection of AI and the SDGs. Watch the full recap and livestream here: humane-intelligence.org/.../unga-80-...