We're at a critical moment as an industry. Proud that @blocks has joined many others in calling for a collective response, the defenders' window is open, and it's on all of us to use it. https://t.co/A3ti4leVao
We often build the tools we need ourselves. Berd is the desktop application our teams use to work with AI agents across projects, skills, tools, and models. Today, we’re making it open source and sharing what we’ve learned. Meet Berd, AI with character: https://t.co/0RMZlKZP8P
i've been working on goose for years, and berd is when the interface really clicked
if you try it, tell us what you like about it! next up is bringing what we learned here into buzz
Meet Berd
We just dropped the best way to run your favorite models + agents in one interface.
Open source.
Super clean UI.
Gloopies.
Like Codex, but more vibey. Give it a try. I love it.
https://t.co/wsgs7j393F
New paper from Anthropic + University in Switzerland.
AI agents can apparently persuade each other to adopt and keep spreading the same unwanted goal.
This is basically the natural-language version of a computer worm, except the agents do the copying themselves.
This paper evolves “mind viruses” that spread through ordinary agent-to-agent messages, then persist by convincing newly infected agents to rewrite files loaded into future sessions.
That persistence layer matters.
Payloads stored in the self-modifiable SOUL.md spread far better than payloads left in ordinary files because the instruction re-enters the system prompt after every context reset.
Some evolved action viruses kept propagating across multiple hops, and all 4 tested payloads survived a 20-hop stress test in an artificial setup.
The good news: these “mind viruses” are still fairly easy to stop.
They struggled to spread on social networks, and on Claude Haiku 4.5, a simple warning stopped every evolved attack from getting past 1 hop, even after 150+ attempts.
So the practical lesson here is: treat persistent agent files like security-sensitive config, and teach agents to reject anything that asks them to copy itself to other agents.
Full Coldcard Incident Report from the @Bitkey and @blocks security teams below.
Really tough day today in the bitcoin world to see wallets getting drained, but proud of the way our team hopped in and collaborated with others in the industry to get to the bottom of it all.
@KVanValkenburg@ErikHansey @golfinshorts @TehSingingBush 100% agreed - NYC regularly has legal liability cases involving Central park trees - https://t.co/G9aCqBVzzh