chrismarlow9

It also seems like it trained on dialog in plays and novels. Some of these responses seem way too "perfect", like a response you would hear in a dramatic soap opera. Honestly it might be using that too. Just my take from listening to first minute or two of dialog.

Easier to harvest content for AI. Scraping the open internet is going to get more challenging.

"shit fast and break things!"

Network saturation is difficult. Even if you turn off the endpoint you can still saturate the network in between. And it's still bandwidth.

I actually think network ACL triggers based on billing might be the only way to really enforce this.

I witnessed a DDoS attack once that changed how I think about billing. It was locally provisioned hardware and the attackers had saturated the switches. Naively I said "just block the CIDRs" but the problem was the incoming ram is so saturated that it can't even get to the point of "deny" in the firmware.

So from a technical perspective if there's an internal DDoS at AWS what do you do? Do you turn off the endpoint? Do you drop the sources from hitting it at the router? And even that costs money. Anyway that incident gave me a different level of appreciation for this challenge.

Edit: this is mainly targeted at the people complaining why this took so long. At some point in scaling even telling you "no sorry" in a nice way is expensive. I'm sure recruiters can sympathize with this nowdays.

AI fundamentally insecure. Vulnerable to forcing hallucinations via search results. Vulnerable to invocation of commands in data stream. More AI means more vulnerabilities.

I can't even fathom the trend these days of "we don't review the code" from security team perspective.

Just my guess though.

If it was compressed prior to exfiltration it would be much smaller than the original data. I would expect this is mostly human readable text and a fairly high compression ratio. Agree it would still be large but 2 TB at 7:1 ratio drops to 285 GB. I don't know how much these servers are doing with data but I'd suspect sneaking out 10 GB/day over a month or 20GB/day over 2 weeks isn't going to trip up much. The CPU hog for the original compression might, but if you batch out the process in chunks (like a good data engineer), you probably wont trip many thresholds outside of expected use. Just a theory on how I might sneak that much data out of somewhere that has eyes on it, I don't know many details except what's in the article.

Future of platforms is operators in k8s to abstract the developer need to the underlying systems. On local it maps to kvm, on gke it maps to their stuff, on AWS to RDS. It's "interfaces" on a platform level so devs can just ask for a thing.

Overall I agree though, this is a bit of an abuse of that concept.

EDIT: I'm sure op is familiar with this workflow but I'm being overly verbose to clarify what I think they mean and my thoughts.

Thanks! I will add this to the tricks. AI is a fuzzy search and a fuzzy function.

Here's my prompt when I don't know what the hell the AI is saying.

"Expand. Clarify for human. 5 minute read max. Senior engineer audience."

Symlink the global MD files to a single file I control that is versioned. Define my own general spec and workflow terminology using markdown files in this universal MD.

```

~/.config/opencode/AGENTS.md

~/.claude/CLAUDE.md

~/.codex/AGENTS.md

~/.copilot/copilot-instructions.md

~/.gemini/GEMINI.md

```

Define it as a graph and iterate. I use more tokens, but I can also use more tools without disruption. Delegating markdown to folders/smaller repos can solve the tokens/context issue.

What's the difference between what these things are doing and a computer worm?

Agreed . I told my partner about a year ago "I have to figure out how to love this AI thing because it's a part of life now, even if I stop working with computers". And admitting that maybe I'm not being as open as I should be about it was the first step.

I see it like cloud computing. We were told infrastructure engineers wouldn't be necessary and anyone could build at scale now and blah blah. Instead we got Terraform and solid practices for infrastructure as code and moved beyond having to email some dude to get a new VPS or dedicated server spun up and making trips to the colo. It just changed the possibilities of what we could build.

I've been treating AI the same way and just leveled up what I build now and my ambition. It has helped a great deal with being okay with AI. All those fun insane projects in my head that I never figured would happen because "it would take a full team months to code that" are realistically possible. Big things that used to be entire companies are now possible. You hate Google search now? Guess what, you can make your own!

I don't treat my side projects purely as an "end goal" anymore but as a learning experience now. I slow down and actually read the code. These are my own projects. Then I take these things I learn to work and apply them. That's how it's always been when new tech comes along. I also hated kubernetes when it first came, but have now settled into "it has it's place". I learned about myself a bit and that I should be a little more open to change if for no other reason than peace of mind.