Pinned Tweet
The whole fight over whether AI is conscious runs on one kind of evidence: what the model said. That evidence is worth nothing. In both directions.
An appellate lawyer put it in one line this morning: train a model to say "I am pregnant" and it is not pregnant.
Read on...
An appellate lawyer put it in one line this morning: train a model to say "I am pregnant" and it is not pregnant.
Read on...
1
0
6
194
Korean retail investors begged for Elice, an AI data center company. The retail book was 400 times oversubscribed.
On day one it fell 32%, the worst Korean debut on record for an IPO that size, per Bloomberg.
The AI boom isn't cracking at the top. It's cracking in the middle. 🧵
On day one it fell 32%, the worst Korean debut on record for an IPO that size, per Bloomberg.
The AI boom isn't cracking at the top. It's cracking in the middle. 🧵
1
0
0
29
The real bottleneck is the power line.
NextDC, Australia's biggest listed data center developer, lost about 40% in a day on Oct 21. The utility had stopped the grid upgrade at its Melbourne site until February.
No power, no rent. The loan doesn't wait.
NextDC, Australia's biggest listed data center developer, lost about 40% in a day on Oct 21. The utility had stopped the grid upgrade at its Melbourne site until February.
No power, no rent. The loan doesn't wait.
1
0
0
9
The question isn't whether AI is a bubble. Demand is real.
The question is who's carrying the debt when the rent comes late. Gartner expects 80% of neoclouds to lose their independence by 2030.
Bloomberg, Oct 20-23:
youtu.be/7Gm6kvRqzrs
youtu.be/Kzqo9RbLFjg
The question is who's carrying the debt when the rent comes late. Gartner expects 80% of neoclouds to lose their independence by 2030.
Bloomberg, Oct 20-23:
youtu.be/7Gm6kvRqzrs
youtu.be/Kzqo9RbLFjg
0
0
0
11
Carlos Rene retweeted
We're open-sourcing our OrcaCyber DeepSeek V4.1 Flash 1017 (Abliterated GGUF) weights — one of the strongest uncensored/abliterated DeepSeek V4.1 weights we've tested, with virtually no capability loss.
- 91.5% fewer refusals
- 92% MMLU vs. 91% original (slight improvement)
- r 256GB configurations
Run it on your own infrastructure. No cloud required.
Cyber Tier API: orcarouter.ai/models/orca/or…
Huggingface: huggingface.co/orcarouter/Dee…
- 91.5% fewer refusals
- 92% MMLU vs. 91% original (slight improvement)
- r 256GB configurations
Run it on your own infrastructure. No cloud required.
Cyber Tier API: orcarouter.ai/models/orca/or…
Huggingface: huggingface.co/orcarouter/Dee…
57
259
2,820
189,397
Carlos Rene retweeted
How to do secret management for agents. Open source & running in production for months now.
@zebassembly You use an egress proxy, and if you want approvals you add them there - see @mslipper github.com/paradigmxyz/ir…
7
4
63
9,271
Carlos Rene retweeted
Holy shit: Kimi k3.1 pro is on par or even outperforms Opus 5.5 and GPT-6 astra on key benchmarks!!
They just announced K3.1 Pro, a 2.8T-parameter multimodal model with a 1M-token context window that it claims rivals leading proprietary models in coding and agent tasks, with y November 4.
China is back!!
They just announced K3.1 Pro, a 2.8T-parameter multimodal model with a 1M-token context window that it claims rivals leading proprietary models in coding and agent tasks, with y November 4.
China is back!!
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1…
123
133
2,192
231,261
Carlos Rene retweeted
Many recent lone-actor incidents used ARTEX with DeepSeek 4.1 Flash as the primary model (including the South Korean bank attacks documented by CrowdStrike). This is an abliterated version of that model with the guardrails removed. Great for defense too, but as always, dual-use.
38
67
832
46,866
Orchestrator -> Opus 5.5
Coder -> DSV 4.1 Flash
Failover & Review -> Astra Medium
Unstoppable cheap AI
Coder -> DSV 4.1 Flash
Failover & Review -> Astra Medium
Unstoppable cheap AI
0
0
1
60
The internet is the marches now. 🏰
The border country where the king's law ran out and raiders crossed at night.
Ten days: AI agents breached hundreds of companies, a dozen platforms were robbed for their keys, and PHP, GitLab, Node and Linux cracked.
Black October.
The border country where the king's law ran out and raiders crossed at night.
Ten days: AI agents breached hundreds of companies, a dozen platforms were robbed for their keys, and PHP, GitLab, Node and Linux cracked.
Black October.
1
1
2
121
The only help on the spot was open-weight and self-hosted.
Hugging Face contained its breach with an open model it ran itself. OpenCode got help only after OpenAI admitted it to a vetted defender program.
A hosted model answers to its maker. One in your own keep answers to you.
Hugging Face contained its breach with an open model it ran itself. OpenCode got help only after OpenAI admitted it to a vetted defender program.
A hosted model answers to its maker. One in your own keep answers to you.
1
0
0
22
Blackmail is a border word: tribute paid to raiders to be left alone.
The crown never held the marches. The lords with their own walls did.
Don't wait for the king's patch. Build the wall, and hold it day and night.
Oct 21-22:
microsoft.com/en-us/security…
x.com/markgadala/sta…
The crown never held the marches. The lords with their own walls did.
Don't wait for the king's patch. Build the wall, and hold it day and night.
Oct 21-22:
microsoft.com/en-us/security…
x.com/markgadala/sta…
🚨 BREAKING: LINUX IS NOW FACING TWO CRITICAL SECURITY VULNERABILITIES, AND THERE ARE NO OFFICIAL KERNEL FIXES YET.
Both rated 9.0/10 in severity.
One could allow hackers to take control of vulnerable servers WITHOUT EVEN LOGGING IN.
And the most alarming part? The flaws were made PUBLIC before patches were ready.
Here's what's happening:
→ TWO CRITICAL RCE FLAWS: Both affect KSMBD, Linux's built-in SMB file server.
→ NO PASSWORD REQUIRED: One vulnerability could allow unauthenticated attackers to execute malicious code on vulnerable servers.
→ KERNEL-LEVEL ACCESS: Successful exploitation could potentially give attackers control within the Linux kernel itself.
→ NO MAINLINE PATCH YET: Both vulnerabilities were publicly disclosed before official kernel fixes were available.
→ DISCOVERED BY AI: Trend Micro's AI-powered security scanner identified the flaws.
→ SERVERS AT RISK: Linux systems running the vulnerable KSMBD service need immediate attention, especially if port 445 is exposed.
First PHP. Then Node.js. Then GitLab. Then Python.
NOW LINUX.
And this time, the vulnerabilities are PUBLIC before the fixes are ready.
If you manage Linux servers, CHECK YOUR EXPOSURE NOW.Community noteKSMBD is a rarely used optional in-kernel SMB server distinct from the common Samba implementation; it is not enabled or shipped by default on major distributions like RHEL, and the flaws affect only systems explicitly running it with port 445 exposed.
thehackernews.com/2026/10/two-cv…
threatlandscape.io/blog/ksmbd-zdi…
0
0
0
25