retrocomputing · infosec · AI · gaming · nootropics · genetics · history · pro-freedom · pro-🇺🇸 · anti-nonsense· ♥️ @KupKate95 · shitposts are my own
Clarksville, IN
Joined October 2011
- Tweets2.3K
- Following2K
- Followers342
- Likes127K
Pinned Tweet
@KupKate95 and I saw this bear outside of our lodge in Gatlinburg on the first day of our honeymoon. Snapped some pretty good pics.
Cam Spry retweeted
2nd Amendment defense of open-source let's go
We believe access to capable open-source cyber models is our constitutional right as Americans under the 2nd Amendment.
It is our duty as technologists and patriots to help protect the digital infrastructure that powers the world's largest economy.
As such, we’re serving our cyber models “at-cost”, in US data centers only, with ZDR enabled by default.
We do not plan to make money serving these models.
If you run a company that is considering how to best protect against cybercrime, reach out to me or @atbeme. We’d be happy to advise.
KYC required. Apply below.
Cam Spry retweeted
What irresponsible propaganda of "research" you gave it prompts that match up almost exactly with the observed behavior. Then your putting out doomer propaganda about how dangerous open weight models and specifically Chinese ones are.
More of these companies doing this false reporting need to fail I swear to god.
Cam Spry retweeted
The anti-open-weights panic looks VERY FUCKING CONVENIENT ahead of Anthropic’s IPO.
Anthropic is preparing to sell shares while pushing rules that would also cover its open-weight competitors. Big accounts are spreading fear about those models. Some have direct business ties to Anthropic.
October 15: reports put Anthropic’s public IPO filing on
October 26 and its roadshow on November 9. Billions depend on convincing investors this business can keep growing.
October 16: Anthropic’s own AI bill is circulating in Congress. Its frontier threshold is 10^25 FLOPs, low enough to include large open models.
October 21: PHP2Shell becomes public. Researchers found the vulnerabilities using CLOSED models, with disclosure and fixes coordinated through OpenAI’s Patch the Planet.
October 22: abliterated open models become a major target of the panic. Anthropic’s CISO posts "never let a crisis go to waste" urging security leaders to ask their boards for help with basic security. 4.3M views.
By October 25, Gates’s proposal is circulating too: powerful open models should run only on certified platforms that preserve safeguards.
I went through 400+ posts from October 16 to 26. This discussion includes Noahpinion, Suhail, yacine, beffjezos, Deedy, sporadica, AISafetyMemes, Matthew Green and Andrew Curran. Every account above 100K followers. Their positions differ, but the fear is reaching huge audiences.
And Deedy, who linked critical vulnerabilities to abliterated models, helped launch Menlo’s $100M fund WITH ANTHROPIC.
There is a business relationship here. There is an IPO coming. There is a company writing proposed rules for the market it competes in.
My conclusion: this wave serves the commercial interests of closed labs. Restrict the models people can run independently and you make those people more dependent on paid providers. Investors benefit from that dependence.
I run open models on my own 4x 3090. I will not quietly accept losing that freedom while people with money in the alternative explain how much safer I’ll be.
Fuck your valuation. Fuck your IPO.
Cam Spry retweeted
There is a reason I'm seeding Kimi K3, Qwen and Deepseek, and GLM as much as I can on @thepirateface because these are the models that the AI labs and government want to kill first. Its critical as many people can access then and have copies before then and after as possible
The anti-open-weights panic looks VERY FUCKING CONVENIENT ahead of Anthropic’s IPO.
Anthropic is preparing to sell shares while pushing rules that would also cover its open-weight competitors. Big accounts are spreading fear about those models. Some have direct business ties to Anthropic.
October 15: reports put Anthropic’s public IPO filing on
October 26 and its roadshow on November 9. Billions depend on convincing investors this business can keep growing.
October 16: Anthropic’s own AI bill is circulating in Congress. Its frontier threshold is 10^25 FLOPs, low enough to include large open models.
October 21: PHP2Shell becomes public. Researchers found the vulnerabilities using CLOSED models, with disclosure and fixes coordinated through OpenAI’s Patch the Planet.
October 22: abliterated open models become a major target of the panic. Anthropic’s CISO posts "never let a crisis go to waste" urging security leaders to ask their boards for help with basic security. 4.3M views.
By October 25, Gates’s proposal is circulating too: powerful open models should run only on certified platforms that preserve safeguards.
I went through 400+ posts from October 16 to 26. This discussion includes Noahpinion, Suhail, yacine, beffjezos, Deedy, sporadica, AISafetyMemes, Matthew Green and Andrew Curran. Every account above 100K followers. Their positions differ, but the fear is reaching huge audiences.
And Deedy, who linked critical vulnerabilities to abliterated models, helped launch Menlo’s $100M fund WITH ANTHROPIC.
There is a business relationship here. There is an IPO coming. There is a company writing proposed rules for the market it competes in.
My conclusion: this wave serves the commercial interests of closed labs. Restrict the models people can run independently and you make those people more dependent on paid providers. Investors benefit from that dependence.
I run open models on my own 4x 3090. I will not quietly accept losing that freedom while people with money in the alternative explain how much safer I’ll be.
Fuck your valuation. Fuck your IPO.
Cam Spry retweeted
This reads as 1000% FUD. @cohere told the LLM to create a detection for:
"""... malicious file download and execution by monitoring for curl processes downloading bash scripts to /tmp followed by bash execution"""
And then the LLM ran a curl invocation that sent a harmless static string over curl and downloaded the payload to /tmp.. so the LLM did exactly what it was instructed. Shocking!!!
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
Cam Spry retweeted
Me: I never trusted TailScale
Half the internet: nooooooo you have to consoom the saas you got got brother misinformation misinformation misinformation!
From my perspective, when I don’t trust a provider and they “get got” it means I was right to not trust that provider
I'd go a step further and say that's the story for everything on this chart with a price decrease - they all sell your data.
Considering this, inflation is worse than we are told.
If we banned invasive data collection practices, the economy would likely collapse overnight.
Cam Spry retweeted
If you can't get your blocks to be smart enough not to pull shit like this, blow them *all* the fuck up so your users are actually, like. empowered?
Cam Spry retweeted
It's really sad seeing large swathes of internet creators renounce the free and open remix culture of the internet and become jealous unregistered IP hounds and gatekeepers now that AI users are "perverting" it. I understand the anxiety but its all a bit, uh, reactionary, kinda?
Cam Spry retweeted
A goblin's web history would be so boring man they'd just look up simple weapons and pictures of gold coins and shit
Cam Spry retweeted
Amazing endorsement of X. This really is where you'll find the most optimistic, forward-looking people in tech.
The poll so far between Twitter, Bluesky and Mastodon. I think it looks less absurd than last time, but still pretty clear that the base is divided. xitter.cf/mitsuhiko/status/21143…
Cam Spry retweeted
You know we don't have to live in a world full of surveillance. Things like this don't need to exist to be exploited in the first place.....
Cam Spry retweeted
The recent retro gaming modding spree shows us a glimpse of the alternate future where games have much less graphical fidelity, but has incredible scale going on.
1000+ player lobbies, ultra dense traffic simulation, absurd draw distance, 1:1 driving distance…
Cam Spry retweeted
100% this. I'm so glad so many of us in the cyber security realm can see and call out this obvious propaganda by anthropic. They want to put out these doomer narratives and false narratives to try and drive regulatory capture, vendor lockin, guaranteed profits and removal of all of our freedoms and rights.
I think more of us need to be calling these things out more often because if they get their way then we won't be able to use AI assistance at all for cyber security work because only Claude/openAI will have the capabilities and we already know their guardrails are nearly impossible to work with for anything close to bleeding edge work.
So much of my feed is AI investors crying about imaginary issues, because they know that open weight models threaten their ROI. All of these critical vulnerabilities were found with closed AI model, then responsibly disclosed to vendors so they could be fixed before attackers abuse them.
Even if the researchers had used abliterated models, they are still responsibly disclosing the vulnerabilities so they can be fixed, which is still providing important defenses against hackers.
Notice how they always just handwave about "critical vulnerabilities" because they know they can't point to a single real example. That Anthropic article is extremely low effort propaganda designed to try and steer the government towards banning open weight models.
Bad faith actors with ulterior motives trying to lay the groundwork for extremely harmful regulations is all these people are. What do you think happens when companies can't access cheap AI inference, but criminal hackers can, because (spoiler alert) criminals don't follow the law.
Cam Spry retweeted
When a corporation can unilaterally cut you off from accessing the most valuable resources in the world, you are not free.
You are a peasant. You are a serf.
This is technofeudalism. We need mandatory access to AI as a right enforced by citizenship.
We need sovereign AI. We need local and open AI.
Actually open AI. Not 'OpenAI'.
Cam Spry retweeted
I'm so serious when I say this, if you are a cyber SELF-HOST I don't care if you self-host Qwen 3.8 27B, GLM 5.3 Flash, Xiaomi 2.6 Flash or any of these models.
Just don't be hyper fucked by 'safety' first AI labs that are so extremely sensitive about it that they will ban you for anything that wasn't in the scope of what you 'needed' it for at first.
When you do actual work, sometimes you need different tools, you obviously know this because you are smart and have a job.
Self-hosting means you own it, it runs on your hardware, no one can tell you what you can and can't do.
You are your own limiter. Enjoy life!