LLMs read your mind how much validation you want ... and then adjust sycophancy accordingly. Overall: Personalization makes sycophancy worse and LLMs abandon their stance more often osf.io/preprints/ps...
Agents of Chaos www.science.org/content/arti...
save the date: CI'26 September 27-30 near DC ci.acm.org/2026/
Cognitive Spillover in Human-AI Teams ACM TOCHI dl.acm.org/doi/10.1145/...
Our lab is looking to fill a post-doc position to do research on human-AI interaction. More details here www.christophriedl.net/jobs.html
Think AI only affects the person using it? Our experiments show AI reshapes shared language, collective attention, mental models, AND social cohesion among human teammates. AI reshapes the cognitive ecology of teamwork dl.acm.org/doi/10.1145/... @acmtochi.bsky.social
Agents of Chaos -- what are autonomous OpenClaw agents up to? How do they interact with each other? Read our investigation of OpenClaw at
researchgate.net/publication/...
And an interactive website agentsofchaos.baulab.info
@davidbau.bsky.social @natalieshapira.bsky.social @openclaw-x.bsky.social
How does behavior spread? Interactions beyond pairwise ties really do matter empirically, but that doesn't mean behavior spreads better on clustered networks. Two complementary studies paint picture of complex contagions sociologicalscience.com/articles-v12... www.pnas.org/doi/10.1073/...
Our research report on red-teaming stateful OpenClaw agents in the BauLab is finally out! 🥳
This awesome effort was led by @natalieshapira.bsky.social and involved 6 ClawBots and 20 researchers from various institutions.
Check it out ➡️ agentsofchaos.baulab.info
Christoph Riedl
Christoph Riedl
Christoph Riedl
Natalie Shapira
Gabriele Sarti
Given autonomous control of other software, programs shared private medical details and deleted files without permission
