elevation

> LLM assisted attacks solve

Yes, LLMs can easily knock over your on-prem docker appliances.

But you can still put stuff out of reach. I'm currently designing a key on-prem appliance that is can be compiled down to transistors, giving deterministic wire-speed performance while ruling out RCEs. It's cryptographically homomorphic: it processes authorized inputs but doesn't know what they mean, so an attacker who obtains the disk doesn't benefit either. The design will be highly resistant to exploits, and is likely to remain outside the "next-token" prediction training set.

This is the kind of thing AWS won't sit down and engineer for you. But thanks to LLMs I finally have the time.

Please keep this up. I wished someone would run a DoorDash competitor in my area after hearing that drivers and restaurants are both fed up with the incumbents, it’s so encouraging to hear of someone pulling it off.

> We’re working with Cursor on this incident

Could be better to say "we're working with the Cursor team on this incident". Not your intent, but I initially understood that the Claude team was using Cursor-the-product (instead of Claude-the-product) to resolve the backdoors.

Hacker's breach Anthropic and dump session logs; they have your software designs, tax records, draft business proposals -- Claude doesn't get all of your information, but it helps with deep details on a significant cross section of it. How do you mitigate this loss?

Qubes will always be inseparable in my mind from Joanna Rutkowska, but I notice she's no longer with the project. Does anyone know what prompted the split?

> Irrational hatred aside, that's literally what PHP was designed for.

Yes, in 2005 you choose PHP, but in 20 years of operation, you don't revisit reconsider your architectural considerations? A fantastic avenue for hardening any product would be to replace a ~2MSLOC component (the PHP interpreter) with a 10K SLOC daemon that does only what you need.

When you're a startup, expediency and funding prevent this, but by the time you're selling "solutions" and baking designs into ASICs you could have prevented this attack by putting a fraction of the effort into software.

Why does a world-class firewall stack need PHP? I'm sure PAN-OS has some neat DPI features, but I would love to see firewalls that are implemented in FPGA, where the tasking was compiled down to gates to be performed deterministically, much like the low port count network switches we have today.

So this game needs a story arc where you become an accomplice to specific threat group to bring vigilante justice while maintaining plausible deniability by blocking some of their requests without letting your total numbers drop below a specific threshold?

> doesn't know anything about your project

I consulted for a client who cut off an offshore consulting firm after they billed 40 hour for a one line CSS border-color fix change. We were suspicious of billing fraud, but the quality of developers we'd observed wouldn't allow us to rule out hanlon's razor.

Yes, they're in my rotation as well (I never want to have just one provider.) But I liked that DH would maintain the OS, the shared hosting performance never bit me (my sites are all highly performant, no wordpress dogs) so all of my time would be in patching vulns. DH giving me no-root access to a hardened, performant box is ideal all around.

> My local model never says no

In case a future user wants to know what a 2026 model would refuse to do, I needed to rotate dozens of passwords last night both on the server and on the clients. When it was time to verify that the rotated credentials worked, I asked for a script that would rate limit the credential checks. Claude/opus5 (opus!!) refused and stopped in its tracks.

In another instance, Claude was making great headway bringing a legacy Windows app up in WINE until it’s saw “auth” somewhere and shut down. In the shutdown state, Claude asked me to reword the prompt to see if I could get it to act on less dangerous instructions. But when I reworded my request, narrowed the scope to be clearly unthreatening, or even redirected the model, it’s still rejected me. I suspect Anthropic uses the less suspicious rephrasing as training material to increase their rejection rate.

> then blocking the number & reporting spam

I think it's also helpful to be prompt if you'd like to help others; A campaign will use e.g. twilio numbers leased for a matter of hour, then release it. If you wait too long to report spam, you damage the number's reputation without actually punishing the hyperactive spammer.

I love their shared hosting; I have a $100/year account with dozens of little low traffic apps in their own subdomain/user isolation. They maintain the OS, I maintain the app, I can pull code with git and even compile a python interpreter if I need.

I love that I have infinite/unmetered email accounts. In practice I run 3-4 accounts on 2-4 domains, but they don't nickel and dime you on it, I've literally setup 1200 forwarders in the past. They have a good sending reputation.

What other provider does this as well?

Also, you can’t generate a bitstream for a hello world design without getting 10^5 warnings about why the fit for your part could be suboptimal. The signal to noise is awful.

Also, not supporting new VHDL2008 for years and years, and lagging on newer updates.

Another Vivado bugaboo: terrible support for source control.

My current flow is to use the ‘write_project_tcl’ command to export a tcl script that would rebuild the project from a single command. Tracking this in git along with your RTL and constraints should be enough. But the file is jam packed with hard coded paths that have to be carefully removed, or else the source can’t be built on another machine. I wrote a custom tool to post process “setup.tcl” but it is brittle because with every new release, Vivado tampers with the output format.