Sabitlenmiş Tweet
My Keepass - Menubar Passwords for macOS
New version 1.3 released!
- NEW: Search field for faster lookups directly on the menu.
- Bug fixes and improvements.
Check it out!
Apple Store: apps.apple.com/us/app/my-keep…
English
Carlos E. Torres
3.5K posts
Carlos E. Torres
@cetorres
💼 Principal Member of Technical Staff at @Oracle (health, mobile, AI assistants) | 🎓 M.S. in Computer Science from @UCCS, AI/ML/NLP/LLMs
🇺🇸🇵🇹🇧🇷 Katılım Eylül 2008
412 Takip Edilen576 Takipçiler
Carlos E. Torres retweetledi
sir...they just dropped a new LLM with 12 MILLION context window and 10x cheaper than Opus 4.7
Alexander Whedon@alex_whedon
Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens - Less than 5% the cost of Opus Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention). Only a small fraction actually matter. @subquadratic finds and focuses only on the ones that do. That's nearly 1,000x less compute and a new way for LLMs to scale.
English
"By design", @Microsoft?
International Cyber Digest@IntCyberDigest
❗️🚨 Microsoft Edge keeps every saved password in process memory as cleartext from the moment it launches. Microsoft's responsed when reported: "by design." All of them. Including credentials for sites you won't open this session. Researcher @L1v1ng0ffTh3L4N tested every major Chromium browser. Edge is the only one that behaves this way. Chrome decrypts credentials on demand, and App-Bound Encryption locks the keys to an authenticated Chrome process so other processes can't reuse them. In Chrome, plaintext surfaces only during autofill or when a password is viewed, making memory scraping far less useful. What makes this extra weird is that Edge still demands re-authentication before revealing those passwords in its Password Manager UI, while the same browser process already holds every one of them in plaintext. In shared environments, this turns into a credential harvest. On a terminal server, an attacker with admin rights can read the memory of every logged-on user process. In the published PoC video, a compromised admin account lifts stored credentials from two other logged-on (and even disconnected) users with Edge running. Microsoft's official response when notified: "by design." The finding was disclosed April 29 at BigBiteOfTech by PaloAltoNtwks Norway, alongside a small educational tool that lets anyone verify the cleartext storage for themselves.
English
"I don’t think it’s fair to say that Claude Desktop installs spyware, but it does open a system up by expanding the attack surface." 🤔 malwarebytes.com/blog/news/2026…
English
@BrianRoemmele I made an interactive visualization: cetorres.github.io/sierpinski-tri…
Source code: github.com/cetorres/sierp…
English
Carlos E. Torres retweetledi
🚨 NATIONAL SECURITY ALERT (TO BE VERIFIED): ALLEGED MASSIVE LEAK OF 251 MILLION CPFs (BRAZIL) 🇧🇷
A catastrophic post has been detected in which a threat actor claims to possess the largest database in Brazilian history, called "MORGUE." The report indicates a massive exfiltration that would exceed even the current living population of the country.
🏢 Allegedly Affected Entity: Gov.br Portal / Brazilian National Registries 🇧🇷
👤 Threat Actor: Buddha
📂 Leak Volume: 251,720,444 records (25.1 GB)
📊 Allegedly Compromised Data (PII):
CPF (National Identification Number), Full Names, and Gender.
Date of birth, Mother's and Father's names.
Death data: Status of death and date of death (which explains why the figure exceeds 212 million living inhabitants).
Race and City/State of birth.
📅 Date of Data: According to the actor, the information corresponds to March 15, 2025.
🔍 Status: Alleged / Not officially verified. Although the actor has published a free sample of 20,000 rows to validate its authenticity.
Monitor:
analyzer.vecert.io
#CyberSecurity #Brazil #GovBr #DataBreach #Buddha #MorgueLeak #CPF #Hacking #InfoSec #VECERT #Cybersecurity #Privacy #BiggestLeak 🇧🇷🛡️💀
English
AI Doesn't Fix Weak Engineering. It Just Speeds It Up.
Read article below.
English
Incident resolved
This incident has been resolved.
Time posted: Apr 15, 17:42 UTC
English
An update has been posted
We have seen success rates for login to Claude.ai, including via Claude Code, stabilize and are working to fully resolve this issue. We will provide an update shortly.
Time posted: Apr 15, 16:29 UTC
English
An update has been posted
The Claude API has fully recovered as of 8:01 PT / 16:01 UTC. We are currently working on mitigating the ongoing errors for Claude AI. Claude Code users who are logged in are still able to use it, but logging in is still broken.
English
Machine Learning Algorithms Explained
🧠 Machine learning algorithms use data patterns to make predictions or decisions without being explicitly programmed for tasks.
🏠 Linear regression models relationships to predict continuous numeric values like future housing prices.
📧 Logistic regression classifies data into categories, commonly used to filter emails as spam or legitimate.
🌳 Decision trees use a flowchart-like structure to make choices based on specific data criteria.
🌲 Random forests combine multiple decision trees to boost accuracy and prevent errors found in single trees.
📏 Support Vector Machines find the best boundary line to cleanly separate different classes of data.
👥 K-Nearest Neighbors classifies items by looking at the labels of their most similar nearby neighbors.
🗂️ K-means clustering groups unlabeled data into distinct clusters based on shared similarities or patterns.
🎲 Naive Bayes uses probability to guess outcomes, highly effective for fast text classification tasks.
🕸️ Neural networks mimic human brain structures, powering complex tasks like image recognition and voice assistants.
🚀 Gradient boosting builds models sequentially, each correcting errors from the previous one for high precision.
📉 Principal Component Analysis simplifies complex datasets by reducing the number of variables while keeping trends.
🎮 Reinforcement learning trains agents by rewarding desired actions in an environment, used for game-playing AI.
🍎 Supervised learning requires labeled data to teach models exactly what output to expect from inputs.
🔍 Unsupervised learning explores raw, unlabeled data to discover hidden structures or natural groupings.
🎬 Recommender systems often use clustering to suggest movies or products similar to your past favorites.
🚨 Anomaly detection finds unusual patterns in data, vital for identifying credit card fraud quickly.
💬 Natural Language Processing algorithms help machines understand, interpret, and generate human language in chatbots.
🧩 Ensemble methods like bagging or boosting combine multiple models to create one stronger, more reliable predictor.
🗺️ Dimensionality reduction helps visualize complex, multi-dimensional data by squashing it down into a simpler format.
English
Oracle's cloud revenue and AI strategy
🚀 Oracle’s cloud infrastructure revenue soared 84%, signaling massive demand for their AI computing power.
📈 This 84% growth rate highlights Oracle's rapid transformation into a leading AI powerhouse.
🤝 Huge AI contracts with firms like OpenAI are driving this massive infrastructure revenue jump.
🏗️ Oracle is investing up to $50 billion this year to expand their data centers.
📋 Remaining performance obligations hit a staggering $553 billion, ensuring future revenue security.
💰 To fund expansion, Oracle is using a mix of debt and equity financing strategies.
🌐 Major tech giants including Meta and NVIDIA are now relying on Oracle’s cloud platform.
🔒 The company’s AI Data Platform allows businesses to reason across private enterprise data securely.
🎯 Oracle raised its 2027 revenue forecast to $90 billion, reflecting deep confidence in growth.
🧐 Investors are closely watching if this heavy infrastructure spending will deliver strong long-term returns.
English
Oracle's stock and growth strategies
🚀 Oracle is aggressively expanding cloud infrastructure to capitalize on the massive global demand for AI services.
📈 Cloud infrastructure revenue recently surged by an impressive 84% year-over-year.
💼 The company holds a record $553 billion in remaining performance obligations.
🏗️ Oracle is investing billions to build out its data center footprint globally.
💳 Management is using debt and equity financing to fund rapid AI infrastructure growth.
✂️ Restructuring plans include significant workforce optimization to streamline internal operations.
🤖 New AI-powered database features are attracting more enterprise-level customers.
💾 The company introduced a 'bring-your-own-chip' model to reduce infrastructure costs.
📦 Supply chain improvements slashed the time from chip delivery to revenue generation.
⚙️ Fusion Agentic Applications across ERP and HCM drive further platform engagement.
📊 Analysts maintain a mostly positive consensus, citing potential long-term undervaluation.
😟 Investor fears revolve around high capital expenditures pressuring near-term free cash flow.
💰 The company maintains a steady dividend to reward long-term shareholders.
🌐 Oracle is positioning itself as a core provider for hyperscalers and AI developers.
🤝 Many contracts are funded upfront by customers to minimize corporate financing needs.
🎯 Management targets 30% to 40% operating margins by 2030 through efficiency.
✨ The 'halo effect' encourages new AI clients to adopt additional Oracle services.
🔄 Aggressive expansion aims to capture the evolving enterprise AI adoption cycle.
🎢 Stock volatility remains high due to market reactions to massive spending plans.
🏛️ The company remains a dominant force in enterprise-grade database management solutions.
English
@Oracle's stock growth strategies
Oracle is aggressively expanding cloud infrastructure to capitalize on the massive global demand for AI services.
Cloud infrastructure revenue recently surged by an impressive 84% year-over-year.
The company holds a record $553 billion in remaining performance obligations.
Oracle is investing billions to build out its data center footprint globally.
Management is using debt and equity financing to fund rapid AI infrastructure growth.
Restructuring plans include significant workforce optimization to streamline internal operations.
New AI-powered database features are attracting more enterprise-level customers.
The company introduced a 'bring-your-own-chip' model to reduce infrastructure costs.
Supply chain improvements slashed the time from chip delivery to revenue generation.
Fusion Agentic Applications across ERP and HCM drive further platform engagement.
Analysts maintain a mostly positive consensus, citing potential long-term undervaluation.
Investor fears revolve around high capital expenditures pressuring near-term free cash flow.
The company maintains a steady dividend to reward long-term shareholders.
Oracle is positioning itself as a core provider for hyperscalers and AI developers.
Many contracts are funded upfront by customers to minimize corporate financing needs.
Management targets 30% to 40% operating margins by 2030 through efficiency.
The 'halo effect' encourages new AI clients to adopt additional Oracle services.
Aggressive expansion aims to capture the evolving enterprise AI adoption cycle.
Stock volatility remains high due to market reactions to massive spending plans.
The company remains a dominant force in enterprise-grade database management solutions.
English