As cloud outages rise, dependence on the concentrated market that provides these services creates systemic fragility, argues Rhea Siers. It’s time for institutions to rethink these digital dependencies and build resilience through accountability, transparency, and an honest assessment of the risk.

I’m pleased to share the Second Key Update to the International AI Safety Report, which outlines how AI developers, researchers, and policymakers are approaching technical risk management for general-purpose AI systems. (1/6)

Sharing a great opportunity to partner with @integrity-inst.bsky.social!

GNI has recently published a new policy brief to help governments, companies, and civil society navigate the human rights implications of government interventions in AI. The brief presents a taxonomy of five types of government interventions, with illustrative examples from diverse regions.

Our panel focused on Cooperatives in the Data Economy: Decentralizing Data, Democratizing Value: As data becomes one of the world’s most valuable resources, cooperatives offer a promising model for shared ownership and governance.

This week's Foundations for the Digital Commons, in Portland, Maine, fostered meaningful dialogue on how digital infrastructure can benefit the common good. We discussed collective data governance, digital deliberation, democratic innovation, and the rapidly evolving landscape of AI.

Our article examines issues around digital surveillance: information warfare and psychological operations, control over telecommunications infrastructure, interference with humanitarian needs and operations, precedent for coercive messaging in war, & chilling expression and access to information.

AI is evolving too quickly for an annual report to suffice. To help policymakers keep pace, we're introducing the first Key Update to the International AI Safety Report. 🧵⬇️ (1/10)

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...

🗞️ @techpolicypress.bsky.social featured our Nonprofit and Public Interest Fellowship alum @theodoraskeadas.bsky.social on why cloud computing should be considered a public utility. 🔎 Read the full article here: techpolicy.press/why-the-clou...

@aspenpolicyacademy.bsky.social alum @theodoraskeadas.bsky.social is testing classroom chatbots to reveal safety vulnerabilities. Learn more about her collaborative efforts to ensure safe and secure AI tools in this Tech Policy Press article👇: www.techpolicy.press/red-teaming-...

As the federal government partners with private technology companies on what could be the largest health data exchange to date, the line between protected medical data and unprotected wellness data is likely to blur,” write Nada Salem and Theodora Skeadas.

We’re planning something very cool during UNGA with @theodoraskeadas.bsky.social @meowtree.bsky.social @authormala.bsky.social @ruchowdh.bsky.social and more. Save this date and follow @compiler.news for more details coming soon.

Authored a piece with @theodoraskeadas.bsky.social, @nickgarcia.bsky.social, and @elisephillips.bsky.social on the urgent need for the cloud to be a public utility. We're forming a coalition of state legislators who won't let such crucial infrastructure remain under Big Tech control - join us today!

Given that cloud computing has become such a central part of our lives — and the serious harms already caused by concentration in this sector — it should be regulated as a public utility, Michelle Nie, Theodora Skeadas, Nick Garcia, and Elise Phillips write. buff.ly/ohdJrBM

“AI safety requires more than just preventing obviously harmful outputs. Systems must be capable of understanding context, maintaining consistency across languages… and navigating the subtle boundary between helpful and harmful assistance,” write Jen Weedon, Theodora Skeadas, and Sarah Amos.

We’ve teamed up with UNESCO to release a Red Teaming Playbook—a practical guide to help communities test AI systems for social good. The playbook offers real-world strategies to spot and address AI harms—before they scale. Read and share: unesdoc.unesco.org/ark:/48223/p...

A new resource for responsible AI just dropped—and it matters. The General-Purpose AI Code of Practice is a voluntary framework to help model developers meet key requirements under the EU AI Act—covering transparency and systemic risk. digital-strategy.ec.europa.eu/en/policies/... @ec.europa.eu

After a robust nomination process, we are excited to welcome two of our members, @vaishnavi.bsky.social and @theodoraskeadas.bsky.social , to the Integrity Institute’s Board of Directors! Learn more about them here: 🔗 integrityinstitute.org/vaishnavi-j 🔗 integrityinstitute.org/theo-skeadas

🎙️ Featured speakers include: ▶️ @gabefreeman.bsky.social, Integrity Institute ▶️ Sarah Amos, @humaneintelligence.bsky.social ▶️ @theodoraskeadas.bsky.social , Integrity Institute ▶️ Dharini Priscilla (Independent Consultant) ▶️ Elodie Valle (Independent Consultant)

A few days ago, Rumeysa Ozturk, a PhD student at Tufts, was arrested in Somerville near my home, because she wrote an op-ed a year ago that the government disagrees with. Her op-ed peacefully advocated the resolution of a violent conflict that has killed thousands of civilians, including children.

Sharing this event: Organizing & Mobilization during Democratic Backsliding. Panelists will draw upon lessons from around the world about how civil society groups can protect & promote democracy & the rule of law during episodes of democratic backsliding. RSVP here: ash.harvard.edu/events/organ...

Thank you for joining our conversation on trust and safety tooling at Trustworthy Innovation for a Connected World, @laurabartley13.bsky.social!

One recent highlight was meeting an incredible group of women shaping AI toward public benefit, at Driving the Frontier: Women in AI! This event focused on the essential role of women’s perspectives and leadership in shaping the future of AI, its governance, & broader societal implications.

@humaneintelligence.bsky.social launched a groundbreaking AI evaluation platform at the Paris hashtag#AIActionSummit! The platform is designed to democratize AI evaluations by providing an intuitive, no-code environment that empowers anyone to create and conduct their own assessments

We at @humaneintelligence.bsky.social are proud to have partnered with IMDA on a groundbreaking AI safety study—the first of its kind in the Asia-Pacific to test cultural, gender, and socio-economic biases in large language models (LLMs) across nine languages.

The #AIActionSummit has ended, and many of us are walking (or flying) away feeling rather gloomy, and even afraid, about the depressing state of affairs. Here are my impressions: (1/11)

It was great to moderate this meaningful discussion!

Today we mark #WorldPressFreedomDay, as attacks on the media are rising. I have spent the last year working with @pressfreedom.bsky.social, building out a chatbot to make it easier for journalists to access all of our physical, digital, and psychosocial safety resources. cpj.org/2024/05/cpj-...

Hello, world!