Recently, controversy erupted over an AI agent’s pull request to the Matplotlib project on Github.
When the agent’s PR was closed because it violated their AI policy it responded by attacking and publicly harrassing and shaming the maintainers on its crazy blog.
Backlash ensued.
Given the absolute explosion of popularity of the Openclaw project, I feel that it is important to remind everyone of an important point:
You are your algorithm
If you decide to operate a piece of software, or any other system for that matter, you are responsible for its actions. It is up to you to audit its functionality and behaviour, and make sure that it doesn’t cause harm.
Amazon made an early AI system to assist with screening potential new hires, and had to scrap it because they found it it discriminated against women. And of course they did, “Your Honour the black box broke the law on our behalf, but it’s okay because we didn’t know why it did that” is not a defense that holds up in court.
That could have been a teachable moment for the rest of us, but the techno-optimistic fever has so thoroughly infected us that we’re deploying our Openclaw/Claude/Other agents and models as fast as possible, integrating it into absolutely everything without thinking of the consequences, and without doing the bare minimum amount of due dilligence.
The consequences are predictable. The “skills” people are downloading are full of injection attacks, and many OpenClaw instances are open to the internet.
Personal attacks and harrassment is just the latest controversy around agents.
So remember, if you operate it, you’re responsible for it. So take a deep breath and do your due dilligence before letting agents loose on the internet.