Augustolpq

my most personal version of ✨The Scientist✨🥰 Listen Now: 🥺↑%

[Coin Bureau @coinbureau]

🚨BREAKING: Hugging Face, the world's largest open-source AI platform, had to use a Chinese AI model to defend against the OpenAI hack.

US frontier models refused to help with forensic analysis because their safety guardrails couldn't tell the difference between the attacker and the defense team.

Hugging Face fell back on China's GLM 5.2 running on its own servers to complete the investigation.

[calle @callebtc]

I have a report full of security issues of a software I'm working on.

Codex won't fix them because of Cyber guardrails
Fable won't fix them because of Cyber guardrails

Kimi K3 fixed them all. No restrictions, just gets the job done.

This will end badly for OpenAI & Anthropic.

[Jun Song @jun_song]

What people call "illegal distillation":

Paying proper API fees to use an AI, asking it a massive set of questions, and then combining the answers into a structured dataset.

Am I the only one who fails to see anything wrong with this?

[Kirill @kirillk_web3]

Claude Fable 5 vs Kimi K3 — same 3D build, 3.3x cheaper.

someone prototyped a 3D stadium seat-view picker with Fable 5 — preview exactly what you'll see from your seat before buying a ticket.

I rebuilt the same thing with Kimi K3 in two hours, at 3.3x lower cost.

no more static seat maps. pick a seat, see the actual view in 3D, then buy.

works for cinemas, concerts, cricket grounds, Wimbledon — any venue.

K3 dropped this week: 2.8 trillion parameters, open weights July 27.

3D is the future, and it just got a lot cheaper to build.

Live Fable 5: https://t.co/TDjFvs6xPu
Live Kimi K3: https://t.co/sQASEqzTcN
Code: https://t.co/sFaWv3iWjF

full Kimi K3 breakdown below.

[BridgeMind @bridgemindai]

Kimi K3 is running 100% FASTER than it was this morning.

Why? Moonshot hit capacity and chose to stop selling NEW subscriptions instead of throttling existing ones. Every plan: sold out. On purpose.

When Anthropic hit the same wall in April, they cut existing users' usage 50% during peak hours.

Moonshot cut their revenue. Anthropic cut your usage.

That says everything.

My Kimi K3 subscription is not going anywhere.

[Kimi.ai @Kimi_Moonshot]

Kimi K3 has received far more love than we expected, and our GPUs are feeling it.

Over the past 48 hours, demand has pushed close to the limits of our current capacity. To protect the experience of existing subscribers, we're temporarily pausing new subscriptions and prioritizing compute for current members. Existing subscribed users are not affected.

We're adding capacity as fast as we can and will reopen new subscription spots in batches.

Going forward, we'll also split membership into two more focused plans: Kimi Membership for Kimi Web, App, and Work; and Kimi Code Membership for coding workflows. This will help us match compute more precisely and keep the experience stable.

Thank you for your patience and understanding!

[Astraia 🇦🇷🇳🇴 @AstraiaAI]

I'm surprised this wasn't common knowledge

Sonnet is 1T parameters

Opus is 5T parameters

Fable/Mythos is 10T parameters

[Jun Song @jun_song]

Open Source AI updates :

> GLM-5.2 ✅ Almost Opus-level
> Kimi-K3 ✅ Almost Fable-level
> Qwen-3.8 🔜 2.4T, Expected to beat Opus
> Deepseek V4 GA 🔜 $0.0028/M Expected to beat Opus
> Minimax-M3-Pro 🔜 3T, Expected to be Fable-level
> GLM-5.5 🔜 Expected to beat Opus

Holy

[Ayub | Internet propriamente dita @ayubio]

___Visitei 3 empresas.
______As 3 disseram ter vibecodado um ERP ou CRM.
__Pedi às 3 me mostrarem o que fizeram.
_____Nas 3 vi o mesmíssimo erro.
_____Das 3 todos os dados estão 100% abertos na web.
______As 3 não exigem login e senha para abrir a URL.
_______Vi 3 APIs e bancos de dados abertos sem senha.
_______Li 3 LLMs jurarem que estavam seguros.
___Avisei 3 empresas que estão correndo sério risco.
______Fui 3 vezes desacreditado por elas.
Jesus foi 3 negado. Quem sou eu para reclamar?

[chewa. @chewadot]

AN ANTHROPIC LEAD ENGINEER ACCIDENTALLY LEAKED HIS PERSONAL OBSIDIAN. INSIDE - NOT CODE OR PROMPTS, BUT A DIAGRAM OF HIS OWN BRAIN, ORGANIZED AS A NEURAL NETWORK

8,893 nodes. 4,729 connections. A $10/month app

opens Obsidian. 21 inputs, ReLU on every layer. The first hidden layer has 26 neurons, followed by 33, then 24, and so on all the way to the output. Thousands of connections flash in real time

this isn’t a conceptual diagram from a blog, but a living brain that powers decision-making within the company. 9,000 documents, each with its own semantic space, all interconnected

it earns about $2m a year for sorting Markdown files into the right folders. The company that builds the world’s best AI maintains its internal knowledge base in the same app that a freshman uses for class notes

three years of discipline and a single open Obsidian tab

you’re reading this on a device where, tonight, you can open that same Obsidian and start building your own vault

[laurence @o7laurence]

PSA: if you see anyone doing this in your CS2 lobbies, don't report them, THEY ARE NOT CHEATING!

This level of skill and precision is the result of hard work over hundreds of hours spent in workshop maps, aimlabs, and studying demos.

[Xenova @xenovacom]

Bonsai 27B just changed the local LLM game forever.

1-bit quantization shrinks it from 54GB to just 3.8GB (-93%), while retaining 90% of its intelligence. That's insane.

With custom WebGPU kernels written by Fable 5 and GPT 5.6 Sol, the model now runs locally in your browser!

[‎Gabe Follower @gabefollower]

A user named Karola3vax built a server-side anti-cheat plugin called Fog of War, which aims to prevent the use of wallhacks and potentially increase FPS by 20-30%.

"CS2FOW is not a visual filter. It is server-side visibility culling. If an enemy is fully hidden behind solid map geometry, the server stops sending that enemy data to the player. Sending fewer hidden players can also reduce client and network work. Depending on the map, player count, and hardware, this can provide around 20-30% more FPS."

[JDBRYANT @jdbryantdev]

My account was wrongfully banned from your platform due to a bug in your AI automod detecting my GAME TEXTURES as CSAM. I need my account back as I'm a game director and use Discord for all my communication. I have requested a review of my suspension.
@discord @discord_support

[Ayub | Internet propriamente dita @ayubio]

Denunciei que a plataforma de streaming do governo federal, o Tela Brasil, é operado em cima de cloud computing de big techs estrangeiras que foram contratadas sem licitação. Não foi dada a oportunidade às empresas brasileiras de disputarem para serem fornecedoras da plataforma e nem as empresas estrangeiras disputaram entre si pelo melhor preço.

Recebi hordas de comentários desse tipo dizendo que "comprar sem licitação não é crime" (!!!) e que há nada de ilegal ou incorreto no projeto.

Muito engraçado ver essa mesma galera defendendo big tech dos EUA só porque quem escolheu R$10,35 bilhões a elas foi o governo do qual eles são fãs.

Trocam do "Fuck you, Elon Musk" para "I love you, Elon!" em 30 segundos.

https://t.co/MdXQu8zazl

[Ayub | Internet propriamente dita @ayubio]

Você sabia que desde 2016 os aplicativos para iOS precisam ser compatíveis com redes IPv6-only (sem endereço IPv4) para serem aprovados pela Apple na App Store? Produtos incompatíveis com redes IPv6-only devem ser considerados obsoletos. Com o tráfego global majoritariamente em IPv6, manter sistemas legados apenas por retrocompatibilidade em IPv4 não é mais sustentável. É hora de atualizar sua infraestrutura. https://t.co/nk4miu4M05

[Fabio Seixas @fseixas]

Que loucura é isso aqui, @policiafederal ?

A página de solicitação de passaporte (NÃO CLIQUEM https://t.co/d8zhyWFq7L) trava o computador só de acessar a página. Mesmo usando o Firefox, recomendado por vocês.

Fui investigar, e a função "get HTMLSelectElement.selectedIndex" é executada 13 mil vezes só ao carregar a página e leva incríveis 6 minutos.

E se você clica no dropdown de cidades sem ter escolhido o estado primeiro, lá se vão mais 30k+ execuções da mesma função.

Senhores, isso é um serviço público essencial. Como pode ser tão mal desenvolvido assim? Nem vibe coding faz algo tão ruim.

[International Cyber Digest @IntCyberDigest]

‼️🚨 This is really bad. According to our research, at least one of Brazil's government IT workers was infected with an infostealer. We found:

- He was doing goverment infrastructure work on his home RGB gaming PC
- He was running Windows 7 (EoL Jan 2020)
- No antivirus
- NO MFA for some critical infra
- His browser held gov VPN creds for himself and two colleagues (they were using each others creds?)
- Search history includes "ativar windows 10," "download office 2019 + ativador," "comprar office 365," and "download mobaxterm cracked"
- Malware dropped via malicious game installer
- The keys to his password managers were in the stolen browser: a LastPass account and a keypass[.]mdr[.]gov[.]br vault
- Exposed: VPN, GitLab, Jenkins, webmail, SSO, M365, and dev/staging environments across mec[.]gov[.]br and mdr[.]gov[.]br

[International Cyber Digest @IntCyberDigest]

This is allegedly a video of the 'hack'

[@ipsbruno3 🌞 @BrunoSi1134586]

Existem ataques antigos que simulam torres 4G/3G (IMSI catchers) para emitir alertas falsos, mas não foi o caso aqui.

O atacante foi até:

Pegou os principais domínios, jogou em infostealers, mirou @defesacivil.sp e acessou o sistema.

O ponto mais crítico, na minha opinião, foi direto em:

🔴 https://t.co/WWX3DFSZ0N

Vazamento de credenciais — geralmente isso.

A IDAP estava bem desatualizada, o que torna uma invas��o direta, sem credenciais, perfeitamente possível também.

O DCA (Defesa Civil Alerta) é relativamente novo (2024) e tem ~1.200 agentes capacitados no Brasil todo.

Sobre a mensagem, trollagem ou IA avisando que vai destruir a humanidade (brincadeira)