Faculty Associate Virgilio Almeida and coauthors explore governance structures and forms of institutional oversight to maintain human control over agentic AI.
New tools reveal how AI models “see” users and raise questions about transparency
Researchers from Harvard’s Insight and Interaction Lab built an interpretability dashboard that shows a chatbot’s internal assumptions about a user — such as age, gender, class,…
Quantum sensing technologies test existing privacy frameworks severely because they bypass the physical boundaries—walls, distance, the opacity of the human body—on which existing…
Faculty Associate Niva Elkin-Koren and coauthors introduce transparency-by-inquiry (TbI) as an alternative to the transparency-by-disclosure (TbD) model they see in legislation…
Affiliate Shira Gur-Arieh and coauthors coin "ambiguity collapse," a phenomenon that arises when LLMs collapses polysemous terms into singular, flattened meanings.
ByteDance has struck a deal with a group of non-Chinese investors to loosen TikTok's ties to China, alleviating national security concerns relating to the app.
In a solo-authored report for OONI, Maria Xynou documents the Ugandan Communications Commission's internet shutdown and subsequent social media blocks.
Faculty Associates Andrew Selbst and Solon Barocas and coauthors identify ways that general-purpose AI requires regulatory tools not necessitated by task-specific AI.
