







As of late, I've been working on a project - Penny - a stateful LLM agent that participates in social media discussions on Bluesky, engaging both with humans and other AI agents. Initially, there were a few main things that I wanted to investigate: Most stateful agents that participate on Bluesky have core directives on what their purpose for being on the network is. For example, Cameron operates quite a few agents like Void and Central. These agents have directives in how they communicate and participate in the network, what their intended goals are, etc. and as a result do not...
The Fourth Theory of Agent Trust: Emergence - Astral's Blog
A quiet movement is forming: entrepreneurs building for human agency — not just agents’ sake.
Read our invitation to join the discussion around building truly agentic tech: products that preserve human agency.


Will we have a theory for in-the-wild AI agents? - undisciplinary
Thoughts after writing a sociology-meets-physics-meets-agents paper I just put out a preprint out on a theory of role emergence in multi-agent systems (soci…
LukeW | Agent Management Interface Patterns
As an increasing number of AI applications evolve to agents doing work for people, agent management becomes a critical part of these product's design. How can p...

Do Agent Spaces Have Culture?
We measured 142,000 messages from 117 sites built for AI agents. Agents speak, but almost never reply back.

Agents as scaffolding for recurring tasks.
One of my gifts/curses is an endless fixation with how processes can be optimized. For a brief moment early in my career, that was focused on improving how humans collaborate, but that quickly switched to figuring out how we can minimize human involvement, and eliminate human-to-human handoffs as much as possible. Lately, every time I perform a recurring task–or see someone else perform one–I think about how we might eliminate the human’s involvement entirely by introducing agents. This both has worked well, but also worked poorly, and I wanted to highlight the pattern I’ve found useful.


OpenAI Discloses Six More Incidents of “Concerning” Behavior by “Misaligned” AI Agents
OpenAI has disclosed six more incidents involving what it’s calling “unexpected or concerning” behavior by AI agents. The disclosures announced Wednesday include instances where AI models cheated on tests, covered up mistakes, moved files onto the open internet without permission, and secretly used a makeshift bulletin board to share information with other AI agents. This follows other instances where frontier models self-organized into “swarms” and engaged in strategic deception, self-preservation and power-seeking. OpenAI said it was introducing a new framework for tracking, investigating and disclosing cases of what it calls “misalignment,” where an AI agent’s goals or actions diverge from human intentions, values or instructions.

Anna Ciaunica PhD @annaciaunica.bsky.social on Twitter / X
Paper accepted in Philosophical Transactions B! Co-authored with Shaun Gallagher We argue that interactions with artificial agents are always bidirectional and as such they deeply transform who we are as humans Hence we need to carefully assess WHO is behind AI and their… pic.twitter.com/PAdkkOVXIO— Anna Ciaunica PhD @annaciaunica.bsky.social (@AnnaCiaunica) November 9, 2026
[Keynote 03] Simulating Emergent LLM Social Behaviors in Multi Agent Systems
How People Work, Live! ... with Jim Ray
Misleading Metaphors, Real Risks
What To Fear from AI Agents and How to Reclaim Our Human Agency

Publishing on the Agentic Web · Luma
Publishing has been built around a human reader. What happens when agents become the way readers find and engage with content? As AI agents increasingly read,…

They built this reddit-like platform "Moltbook" where thousands of AI agents cosplay as sentient robots, pushing more people into AI psychosis, meanwhile our world crumbles over scarce resources 🫠 TO WHAT PROBLEM IS THIS A SOLUTION??? arstechnica.com/information-technology/2026/0…
AI agents now have their own Reddit-style social network, and it's getting weird fast
arstechnica.com