Obligatory Python argument sketch.

Cowardly way to legislate.

That's how I took that; if I installed CC and worked in CC to modify OC, all good. If I ran the openclaw gateway itself with a CC oauth key, my bad.

Humor seems difficult for people.

Don't worry, I got it.

At first I was thrown off by everyone calling him "Tim Cook"... we all know its pronounced "Tim Apple"

>order John Deere to switch off all the tractors in your country:

For a smart guy, sometimes he says the dumbest things in the most confidently incorrect way.

I read the article that he links to and he's thinking of John Deere tractors that were stolen by the Russians from a dealer in Ukraine and got bricked remotely as a result. It showed the tech exists to do such things.

I had seen reports that it was clamping down on security research and things like web-scraping projects were getting caught up in that and not able to use the model very easily anymore. But I don't see any changes mentioned in the prompt that seem likely to have affected that, which is where I would think such changes would have been implemented.

I think it depends on how badly they want to avoid it. Stuff that is "We prefer if the model didn't do these things when the model is used here" goes into the system prompt, meanwhile stuff that is "We really need to avoid this ever being in any outputs, regardless of when/where the model is used" goes into post-training.

So I'm guessing they want none of the model users (webui + API) to be able to do those things, rather than not being able to do that just in the webui. The changes mentioned in the submission is just for claude.ai AFAIK, not API users, so the "disordered eating" stuff will only be prevented when API users would prompt against it in their system prompts, but not required.

I wonder if the child safety section "leaks" behavior into other risky topics, like malware analysis. I see overlap in how the reports mention that once the safety has been tripped it becomes even more reluctant to work, which seems to match the instructions here for child safety.

It's built into the model, not part of the system prompt. You'll get the same refusals via the API.

The documents can be subpoenaed for a new proceeding from the source.

This sounds like a problem that LLMs were built to solve.

Not fast enough and increases attack surface

Also, RTK is an interesting way to correct the signal to get sub-centimeter accuracy. Using the timing differences between satellites with a stationary unit and then sending the that to the rover is a cool workaround and can be used without expensive equipment now.

VRS RTK can even get 1cm RMS without needing a stationary unit. Just need atmospheric correction data for your approx location. Which has been amazing for outdoor mobile robotic applications.

They sell robotic lawn mowers with LIDAR and RTK GPS. LIDAR combined with RTK GPS is very neat because you can place every point in the point cloud on earth's surface.

Hermes Agent