netsec_burn

If you're responding to the parent post, as most of us are, the Lean proof is not irrelevant.

Verification is equally important, if not more so.

The OpenAI status page is still yellow. Like most modern status pages, yellow denotes the servers are on fire. Red denotes Sam Altman is bleeding out somewhere on the floor, the feds are about to bust in and shut down the GPUs.

Perhaps you meant to reply to a different post?

After using Claude for a long time, I tested Sol 5.6 for the first time today. Love it, its an incredibly capable model and uses far fewer tokens/time thinking. Its what I imagine Fable would be if I haven't been downgraded on every conversation - even after completing the verification program. I think I may cancel my Claude subscription finally.

Addressed in the article, in case you're curious.

Speaking from experience, the LLM agents adapt fairly well to these contexts too. It's not at all FUD, you're at a significant disadvantage if you don't compete with AI now. I went to a CTF recently against teams I have won against every year, and within 10 minutes of the event starting they had solved every challenge. They have an agent loop and it solves everything immediately, so they won. Anyone attempting to solve the challenges on their own has no chance, even if you think "maybe this is too out of the box for LLMs". Furthermore, the DEFCON CTF you're referring to has quals, and if you don't qualify you don't get those challenges in the finals. Quals has mainly binary exploitation challenges which Opus (and others) solve as long as you hold the gas pedal down on your API bill. I don't believe it's hyperbole to say CTF is dead, as a competitor.

I can't even sign up for Max (last tried yesterday), their credit card processor has issues.

Another trope: longer README.md's than anyone would make, or want.

I prefer the Dell Rugged line or Thinkpads, since a single water droplet on the keyboard is enough to kill this laptop.

Also the Kiisu which is 1:1 with capabilities, plus a few more.

The allowed types are a bit misleading. Any binary is accepted, any architecture. You can upload shared objects, ELF executables, PE binaries, etc.

I like to give it bomb executables (reverse engineering challenges) to test it.

I made a site to use LLMs to help me with reverse engineering. The output is surprisingly readable, even with C++ classes. Let me know any feedback you might have: https://decompiler.zeroday.engineering/

I can answer the writing to /proc one. It is sometimes useful to hotpatch running programs with /proc/pid/mem.

Or the LilyGO T-Embed CC1101.

I recently learned this too, just a few months ago. Ended up making a frontend so I could do it automatically: https://decompiler.zeroday.engineering/

I've used wormhole once to move a 70 GB file. Couldn't possibly do that before. And yes, I know I used the bandwidth of the relay server, I donated to Debian immediately afterwards (they run the relay for the version in the apt package).

Not yet, I'm writing it. Will be available by the end of the month, feel free to follow the PR: https://github.com/flipperdevices/flipperzero-firmware/pull/...

How? They are prohibited from using it in the license.

Today appears to be the day you can run an LLM that is competitive with GPT-4o at home with the right hardware. Incredible for progress and advancement of the technology.

Statement from Mark: https://about.fb.com/news/2024/07/open-source-ai-is-the-path...

I fixed this by disabling the Photos app and using Google Gallery (on the Play store). It's the same thing as Photos for what I was using it for, without the online features.

Meta, OpenAI, and Anthropic come to mind.

In TFA they name OpenAI, Google DeepMind and Anthropic.

Hit or miss, with significantly more hits than misses. I was able to build a useful cybersecurity service in about 6 hours leveraging Claude to write most of the code. It has certainly made me more productive.

This more likely has to do with context length?

https://openai.com/index/openai-board-forms-safety-and-secur... (May 28th)

> OpenAI has recently begun training its next frontier model and we anticipate the resulting systems to bring us to the next level of capabilities on our path to AGI.

Have you tried 3 Opus or 3.5 Sonnet? Are you using it for programming, or something else?

Opus remained better than GPT for me, even after the release of GPT-4o. VERY happy to see an even further improvement beyond that, Claude is a terrific product and given the news that GPT-5 only began its training several weeks ago I don't see any situation where Anthropic is dethroned in the near term. There are only two parts of Anthropic's offering I'm not a fan of:

- Lack of conversation sharing: I had a conversation with Claude where I asked it to reverse engineer some assembly code and it did it perfectly on the first try. I was stunned, GPT had failed for days. I wanted to share the conversation with others but there's no way provided like GPT, and no way to even print the conversation because it cuts off on the browser (tested on Firefox).

- No Android app. They're working on this but for now, there's only an iOS app. No expected ETA shared, I've been on the waitlist.

I feel like both of these are relatively basic feature requests for a company of Anthropic's size, yet it has been months with no solution in sight. I love the models, please give me a better way of accessing them.