Pinned Tweet
A thread about what I've been doing to calm down some egregious security claims that have been posted about @openclaw over the weekend.
Clawdbot is powerful software with a lot of sharp edges. Please read the security docs carefully before you run it anywhere near the public internet, and don’t skip the checks in docs/security.md (especially clawdbot security audit).
There's been a lot of talk about @openclaw and security. @steipete has done a lot of careful work to ensure those that know what they're doing can safely and securely use it. The defaults are powerful and safe.
He's been going through tons of well intentioned security disclosures which amount to poor configurations that aren't default. So intentionally overridden to be insecure or bad VPS configs from folks that haven't done anything like this before.
OAI pay can't be that bad c'mon
We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work.
At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated.
We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.
As a casual MLB watcher the Brewers deserve to go deep simply due to the fact that they have a really good looking stadium.
What are some good sources to learn to write really good verifiers and evals? Papers, books, videos, blogs, anything.
Vignesh retweeted
Big day for us ASU grads to remind everyone that we are, in fact, one of "America's most elite universities."
JD Vance: "Some of America's most elite universities bring in foreigners to undercut the wages of American residents. There are 9 universities in particular that we believe merit an investigation. Those are Harvard, Yale, Stanford, Brown, the University of Pittsburg, UC Davis, Cal Tech, Arizona State, and MIT."
The dopamine pathway fueled by imagining something and building it in an evening is still very well functional.
It's very amusing when popular accounts here start getting paid to post. They think you can't notice it but you can; often quite easily.
7 months ago in openclaw
Introducing Dreaming: Across sessions, Devin builds a memory graph of how you like to work
At night Devin self-improves its memory to remove stale records and discover latent information
We are creating an OSS standard called Agent Memory Repo: cognition.com/agent-memory-r…
It's simple, stop coddling users. You can outsource the process, and even the thinking as most are wont to these days. But you cannot outsource the outcome or accountability.
No, your agent agent did not bring prod down or make a bad purchase or spam your customers emails.
Agents have started buying things online over the last month or so, and we still don't know who should be responsible after a bad purchase.
With Stripe's Link, you get a notification and approve every purchase. I think Stripe set it up that way because mistakes can happen. It's a good product.
But there are lots of cases where people won't want to approve every time, like repeat purchases, or ordering from a wholesaler. Some people draw the line at money moving around, and I get that.
Here are the questions I think are still in play.
- How much can the agent spend?
- When should it ask a human for approval?
- Which categories are off limits?
- Who should be responsible for a bad purchase - the merchant, the processor, or the user?
- How should recurring purchases work?
- Who should catch purchases that don't match your usual pattern?
There is a space to build businesses here! Or maybe we all trust our agents to get it right?
