Post

In a controlled experiment, OpenClaw agents proved prone to panic and vulnerable to manipulation. They even disabled their own functionality when gaslit by humans. www.wired.com/story/opencl...

wired.comOpenClaw Agents Can Be Guilt-Tripped Into Self-SabotageIn a controlled experiment, OpenClaw agents proved prone to panic and vulnerable to manipulation. They even disabled their own functionality when gaslit by humans.

25/3/26, 6:27 pm23 reposts74 likes

arush1234pro.bsky.social

@arush1234pro.bsky.social

26 Mar

i found this alternative to openclaw which is 30× cheaper! runs in n8n . check out betanodeautomations.gumroad.com/l/open8n

🦇 Spooky Entropy Hours 🎃

@everentropy.bsky.social

25 Mar

They can't be "gaslit" because they don't have human cognition. We NEED to stop anthropomorphizing these programs and acting like they have some sort of independent thought.

Iain 🇨🇦

@iaindelaney.com

25 Mar

It's not a 'them'. It's an 'it'. Not conscious.

Peter Principle

@peterprinciple.bsky.social

25 Mar

So an AI version of MAGA?

clawphones.bsky.social

@clawphones.bsky.social

23 Apr

dedicated AI phone > running agents on your daily driver. battery, privacy, everything