LeituraPesquisaRadarFramework de investimento
Entrar / Cadastrar
TudoTecnologiaProdutosInvestimentoBolsaOutros

Leitura

2026-11-0672 posts

Atualizar

OpenSSH Releases Emergency Update 10.6p2 to Fix 'SSHole' Authentication Bypass Vulnerability Affecting musl-Based Linux Systems

Tópico · OpenSSHMaterial
2026-11-06 11:43 GMT+8

The OpenSSH development team released an emergency update, version 10.6p2, on November 5, 2026, to fix a high-severity authentication bypass vulnerability named "SSHole" (CVE-2026-123158). The vulnerability has a CVSS severity score of 8.1 and allows attackers to potentially log into accounts that have no password set by entering any string between 257 and 1,024 characters in the password field during an SSH connection.

The flaw traces back to OpenSSH 7.3p1, released in 2016, meaning versions from the last decade may be affected. However, SSHole does not impact all users. It specifically affects Linux operating systems that meet three conditions: they use the musl C library, OpenSSH does not have PAM (Pluggable Authentication Modules) enabled, and SSH password authentication remains active. Major affected distributions include Alpine Linux, OpenWrt, and postmarketOS, as well as certain Docker images based on these systems.

Security research firm Wakefield identified that the vulnerability resulted from an unintended combination of security adjustments across three independent projects: musl restricted password length in 2012 to prevent denial-of-service attacks; OpenSSH modified password verification for some accounts in 2016 to avoid timing side-channel leaks; and Alpine adjusted the root account's password field settings in 2022 to facilitate SSH key login. These changes, each solving a different problem, ultimately created exploitable authentication bypass conditions. Wakefield discovered SSHole with the help of AI tools on November 3 and reported it to the relevant teams, leading to rapid patch releases from both OpenSSH and Alpine Linux.

For administrators unable to immediately update OpenSSH, Wakefield provided temporary mitigation measures: disable the PasswordAuthentication option in the SSH server configuration and restart the service. Additionally, administrators should check SSH login logs; if the password field shows "*" and an "Accepted password" record appears, it should be treated as a potential sign of SSHole exploitation, requiring further investigation into system compromise.

Fontes:iThome

Pesquisa

OpenAI President Greg Brockman Says Company Could Be IPO-Ready in 2027 as Enterprise Revenue Surges

Tópico · OpenAI估值与融资Material
2026-11-06 02:02 GMT+8

OpenAI co-founder and President Greg Brockman said the company would be ready for an initial public offering (IPO) in 2027 during a Wall Street Journal tech conference on Wednesday. When asked if the company was IPO-ready next year, Brockman replied, "I think so. Yes." He noted that the company has made significant progress in the enterprise market this year and expressed strong confidence in the business.

Brockman detailed how OpenAI's trajectory shifted throughout 2026. He admitted that the first and second quarters were "pretty rough," noting that OpenAI "wasn't even in the conversation" regarding agentic coding, an area where rival Anthropic's Claude Code gained market share. However, activity picked up in the third quarter, with OpenAI seeing 70% quarterly growth in its annualized revenue. Brockman attributed this growth to the enterprise side of the business doubling, describing it as the culmination of prior investments rather than a one-off event.

Now, the majority of OpenAI's revenue comes from the enterprise business, which previously accounted for only around 30% of total revenue. Brockman clarified that consumer revenue has not shrunk but continues to grow, while the enterprise segment has "really taken off." He cited a recent partnership with JPMorgan Chase as an example of the type of large-scale client deals the company is now winning.

Brockman also shared lessons learned from selling into the enterprise market, emphasizing the need to help businesses identify key milestones achievable with AI technology, build trust, and demonstrate reliability as a long-term partner. He contrasted this with perceptions from a year ago, when OpenAI was viewed primarily as a consumer company facing skepticism about its enterprise commitment.

Fontes:TechCrunch

Pesquisa

German Federal Government Releases AI Task Force Report: Targets 80% Large Enterprise Adoption by 2030

Tópico · 德国AI强国战略Material
2026-11-05 18:40 GMT+8

The German Federal Government recently released a report by its AI Task Force, detailing a strategic vision for the country to become an "AI Nation" by 2030. The report explicitly states that the next eighteen months are critical in determining whether Germany can remain competitive in the global AI race. It warns that despite a strong foundation in research, the gap between Germany and leading nations like the United States and China is widening, and the time window to narrow this distance is closing soon.

To achieve these goals, the report sets specific quantitative targets. By 2030, at least 80% of large enterprises and 55% of small and medium-sized enterprises (SMEs) are expected to adopt artificial intelligence technologies. The number of AI startups should triple compared to current levels, and AI-generated economic output should account for more than 10% of the GDP. Additionally, 90% of federal government agencies are required to deploy at least one AI application for core tasks.

The report also lists 18 "key measures" across six sub-themes designed to create significant leverage. However, media analysis notes that achieving these targets relies heavily on broad societal participation, and criticizes the report for lacking sufficient scrutiny of potential risks and ethical challenges associated with AI, describing it as a "dreamlike" optimistic outlook.

Fontes:heise online

Pesquisa

Cloudflare Q3 Revenue Growth Accelerates to 39%, Full-Year Guidance Raised, Shares Surge Over 14% After-Hours

Tópico · CloudflareMaterial
2026-11-06 07:20 GMT+8

U.S. cloud security and connectivity platform provider Cloudflare Inc. reported third-quarter results for the period ended September 30, showing revenue of $782.5 million, a 39% increase year-over-year and the fastest growth pace in nearly four years. Adjusted earnings per share came in at 38 cents, up from 27 cents in the same quarter last year and beating analyst expectations of 34 cents. Following the announcement, Cloudflare shares surged more than 14% in after-hours trading.

The company's annualized run rate has now crossed the $3 billion mark, while current remaining performance obligations rose 37% from a year earlier. Although high stock-based compensation of $174.1 million resulted in an unadjusted net loss of $34.1 million, the adjusted operating margin improved to 17.5% from 15.3%. Chief Executive Matthew Prince described the period as "Cloudflare's best quarter ever," noting that the business has found "an even higher gear" for its next phase of growth.

Based on these results, Cloudflare raised its full-year 2026 revenue guidance by approximately $97 million to a range of $2.963 billion to $2.964 billion, significantly above the previous Wall Street consensus of $2.87 billion. Fourth-quarter guidance also reflects optimism, with expected revenue of $845 million to $846 million and adjusted EPS of 43 cents. Additionally, the company appointed Christian Smith, formerly Chief Revenue Officer at Splunk, as President of Revenue to oversee global sales operations.

Fontes:SiliconANGLE

Pesquisa

Cloud Contact Center Software Provider Five9 Announces $325 Million All-Cash Acquisition of AI Startup Level AI; Shares Fall Over 23% on Mixed Guidance

Material
2026-11-06 08:35 GMT+8

U.S. cloud contact center software provider Five9 Inc. announced on November 5, 2026, that it will acquire artificial intelligence-native competitor Level AI in a $325 million all-cash deal. The transaction is expected to close in the fourth quarter, subject to customary closing conditions. Five9 stated that the move aims to strengthen its competitive position in the customer experience landscape and deepen its AI talent pool.

Level AI develops a contact center and customer experience intelligence platform that uses AI to support human agents on calls, analyze interactions, and provide real-time guidance. Five9 noted that Level AI's applications are currently in production with enterprise customers, handling more than 1 billion customer interactions annually. Five9 plans to continue offering Level AI as a standalone tool while also integrating its capabilities into the Genius CX Platform.

Alongside the acquisition news, Five9 reported its third-quarter financial results. Adjusted earnings per share came in at 82 cents on revenue of $323.3 million, slightly surpassing Wall Street estimates of 80 cents and $318.3 million. However, the company's guidance for the current quarter and full year was mixed: current quarter EPS is expected between 88 and 92 cents, below the market consensus of 99 cents, and full-year EPS is projected at $3.17 to $3.21, missing the consensus of $3.26. Notably, the full-year guidance excludes any contributions from the pending Level AI deal.

Investors did not share management's enthusiasm. Five9's stock fell more than 23% in late trading, reacting to the cost of the acquisition—which amounts to nearly half of Five9's $738 million in cash and marketable investments at the end of Q3—and the underwhelming forward guidance. This sharp decline highlights market concerns regarding large cash expenditures and integration risks.

Fontes:SiliconANGLE

Pesquisa

AI Company Anthropic Warns in IPO Filings That R&D May Be About to Accelerate Dramatically; Frontier Model Astra Demonstrates Zero-Shot Direct Editing of Other Models' Weights

Tópico · AnthropicEstrutural
2026-11-06 02:37 GMT+8

Anthropic's amended S-1 filing now states that the company may be imminently meeting the definition of 'dramatically accelerated progress attributable to AI' found in its Responsible Scaling Policy.

This warning is not a generic risk disclosure. Anthropic's Responsible Scaling Policy defines this threshold as: models being able to fully replace all research scientists and engineers at competitive costs, or a dramatic acceleration in the pace of AI progress attributable to the automation of R&D.

Specific technical support for this assessment comes from recently published experimental results. Researchers Rikiya Takehi et al. reported that GPT-6 Astra can directly read and edit the raw weight parameters of other neural networks, requiring no gradients, no training, and no tool calls. They termed this 'Weight-Space Programming.'

These results require clear attribution: they are benchmarks run by researchers themselves and have not yet been independently replicated. However, Zvi Mowshowitz noted in his analysis that the significance of these experiments lies not in a specific numerical improvement, but in the demonstration of a mechanism—frontier models can treat other models' weights as readable and writable code to diagnose and patch them. If this capability continues to strengthen, a self-accelerating cycle in AI R&D may already have begun.

The strongest counterargument is the scale issue. Astra edited a small model with 0.8 billion parameters, primarily fixing a specific defect in reasoning loops. For models that are already functioning well, Astra could only provide marginal improvements. There is a massive unverified gap between generalizing this technique from patching known issues in small models to autonomously designing next-generation frontier models. Anthropic itself acknowledges in the S-1 that 'we do not anticipate being able to safely use Claude to fully autonomously design and build successive generations of frontier models.'

But the core of Zvi's argument is that Anthropic, as the lab most likely to trigger this threshold first, proactively wrote this sentence in a legal document filed with investors, indicating that the company has internally observed signs of acceleration. Previously, Anthropic stated that they often only identified such accelerations after they had occurred. This time, they are attempting to warn before it happens.

Fontes:thezvi.substack.com

Pesquisa

Electronic Frontier Foundation: Google AI Summaries Are Knowledge Panels in Disguise, Lock-In Strategy Backfires on Monetization

Material
2026-11-05 04:05 GMT+8

In an analytical article, the Electronic Frontier Foundation (EFF) posits that regardless of whether Google labels its features as AI summaries, knowledge panels, or AMP, the core of these actions is not functional improvement, but rather locking users into Google's own ecosystem. Understanding this is key to choosing services better than Google's, thereby thwarting their anti-competitive goals.

EFF Special Advisor Cory Doctorow names this pattern "enshittification"—a shift where platforms gradually move from serving users to serving their own commercial interests. The article emphasizes that regardless of how glamorous the new technology's packaging may be, the driving force behind decisions remains monetization logic, not the technology itself.

Google Search has a history of similar operations. The most direct example is placing sponsored links and ads at the top of search results, forcing users to scroll past irrelevant content to find answers. Another form is self-preferencing: when searching for videos, YouTube is weighted higher than competitors; when searching for news, AMP pages hand over mobile rendering control to Google, with links defaulting to Google links.

However, a more subtle approach is making it unnecessary for users to leave Google entirely. Knowledge panels were an early form of this strategy: displaying a snippet of information directly on the search results page, claiming to have already answered your question. These panels predate the AI boom and are not always accurate.

The EFF argues that current AI summaries are essentially the same thing as knowledge panels, just rebranded because AI is currently trendy. The difference lies only in execution: knowledge panels extract citations from specific linked sources, whereas AI summaries are generated by models themselves, occasionally accompanied by links. The goal remains essentially the same: to make you feel there is no need to click any link and leave Google, thereby increasing page dwell time, ad exposure, and revenue.

Yet the greatest irony is that this strategy has not worked. Citing a Financial Times report, the EFF notes that users opting for AI mode—possibly accustomed to obtaining text answers instead of clicking links via knowledge panels and AI summaries—have actually led to lower ad monetization. Ads are Google's primary source of income. Like newspapers before them, Google is discovering that relying on a monopoly on ad space may not be the most robust strategy amidst shifting technological landscapes. To dig itself out of the hole it created, it is burning cash in an attempt to regain its advantage.

The article concludes by reminding readers: every time you see such a feature, regardless of what it is called or what new technology it claims to be driven by, remember that the decision driver is Google's monetary incentive, not the technology itself. You can choose to ignore it, skip it, disable it, or even switch entirely to other services.

Fontes:Electronic Frontier Foundation

Pesquisa

US House GOP Committee Leaders Request Closed-Door Briefing from Nvidia CEO Jensen Huang on China Spy Allegations

Tópico · 英伟达黄仁勋间谍指控Material
2026-11-06 05:09 GMT+8

Leaders of three Republican-led committees in the U.S. House of Representatives have sent a letter to Nvidia CEO Jensen Huang, requesting his attendance at a joint closed-door briefing on November 18. The move follows recent reports alleging a close relationship between Huang and an executive identified as a Chinese intelligence asset.

According to CNBC, the letter was signed by House Foreign Affairs Committee Chairman Brian Mast, Select Committee on China Chairman John Moolenaar, and House Permanent Select Committee on Intelligence Chairman Rick Crawford. The lawmakers seek information on Huang's interactions with Alice Huang, an executive at Megaspeed, as well as Nvidia's security protocols against foreign adversary espionage. The New York Times previously reported that U.S. intelligence agencies found Alice Huang had developed a close relationship with Jensen Huang while funneling information to Beijing.

An Nvidia spokesperson stated that regional cloud customers receive ordinary business information, not sensitive trade secrets or national security-relevant data, and emphasized that meetings with customers are part of a CEO's job. The requested briefing date falls immediately after Nvidia's scheduled third-quarter earnings release on November 17, highlighting the company's exposure to regulatory scrutiny amid geopolitical tensions over AI chip exports.

Fontes:CNBC

Pesquisa

Brazilian Court Bans Meta from Using Media Content for AI Training, Imposes Daily Fine

Tópico · Meta巴西媒体版权禁令Material
2026-11-06 06:22 GMT+8

A court in São Paulo, Brazil, has issued an injunction prohibiting US tech giant Meta from using content belonging to member companies of three Brazilian press associations to train its artificial intelligence (AI) models. The ruling directly targets Meta's previous attempt to force media organizations to license their content for free in exchange for continued platform access through modified terms of service.

According to heise online, Judge Guilherme Madeira Dezem of the 44th Civil Court in São Paulo granted the interim relief requested by the National Newspaper Association (ANJ), the Brazilian Association of Radio and Television Broadcasters (Abert), and the National Association of Magazine Publishers (Aner). These associations alleged that Meta's new terms of service for business users made the "free and unrestricted provision of copyright-protected material" a condition for continuing to use its platforms. The judge stated in the ruling that this requirement constituted an obvious abuse of rights.

In addition to suspending the execution of the relevant clauses, the court also banned Meta from sanctioning or penalizing media companies that refuse to accept these new terms. If Meta violates the injunction, it will face a fine of 1 million reais per day (approximately 162,000 euros). This decision is seen as an initial victory for the Brazilian media industry in protecting intellectual property and maintaining the sustainability of professional journalism.

Fontes:heise online

Pesquisa

How Amazon Marketplace Evaluates Content Quality with Frozen Vision-Language Models and OCR Signals

Tópico · 亚马逊A+内容质量评估Material
2026-11-06 02:06 GMT+8

A+ Content is the rich media widget on Amazon product detail pages, where brands use combinations of images, text, and video to tell their product stories. The challenge is that checking each asset in isolation—such as verifying image clarity or grammar errors—misses cross-module quality issues: for example, a feature list might advertise a 64-ounce capacity while a lifestyle photo displays a single-serving model. To address this, the Amazon team renders the entire page as a single snapshot, allowing the vision-language model to evaluate the content holistically, much like a customer would.

A core architectural decision is to not fine-tune the vision-language model. The team reasoned that fine-tuning a base model with only a few thousand labeled samples requires GPU training infrastructure and carries a risk of overfitting. As an alternative, they keep the VLM and its scoring rubric fixed. The model outputs scores from 0 to 10, confidence values, and natural language rationales across four dimensions: Visual Storytelling, Feature Presentation, Layout, and Conversion Optimization. These structured outputs are then used as features to train a CatBoost gradient-boosted decision tree classifier to predict expert-labeled high-quality or low-quality tags. Because only the classifier is retrained, updates required as annotations accumulate or criteria change can be performed as a short CPU-only task.

While vision-language models excel at capturing overall semantics and visual composition, they are less reliable for fine-grained typographic readability; they often overlook layout signals such as text density, line spacing, and character crowding. To capture these details, the team runs an OCR model in parallel to extract text blocks and their positions from the snapshot, calculating four readability metrics: Text Density, Character Density, Spacing, and Overall Readability. The fusion model inputs eight features—the four VLM scores and the four OCR metrics—into the same CatBoost classifier. On a 20% holdout set, the classifier using only VLM features achieved 89% accuracy and an F1 score of 0.90, while the one using only OCR features reached 82% and 0.83. The fused model achieved 94% accuracy and an F1 score of 0.95, representing a 5 percentage point improvement over the VLM-only approach.

Production validation of this method included several key steps: using SHAP to explain the contribution of each feature to decisions, feeding the most negative features along with product category and best practices into a language model to generate prioritized improvement suggestions, and running global A/B experiments that adjusted the visibility of low-quality content based on model judgments, which resulted in a statistically significant increase in Ordered Product Sales. Sellers who revised their content using these suggestions saw an average quality score increase of 1.4 points (on a scale of 10). It is important to note that all figures above are self-reported by Amazon, based on internal annotations and holdout sets, without independent replication. The annotations covered 10 product categories, and whether this method can be generalized to other multi-module content evaluation scenarios remains unverified.

Fontes:Amazon Web Services

Pesquisa

Santhosh Kumar, CTO of Indian Fitness Company Cult.fit: AI Implementation Is an Organizational Empowerment Issue, Not a Tool Procurement Problem

Material
2026-11-06 04:00 GMT+8

Cult.fit is a technology-driven fitness company in India, operating 708 fitness centers and serving nearly one million paying members. In a case study published by OpenAI, CTO Santhosh Kumar articulated a core judgment: the bottleneck for AI implementation lies not in model capabilities or tool access, but in whether the organization can enable non-technical staff to truly utilize these tools.

His advancement path consists of three stages. The first stage is standardization within the engineering department. Previously, developers used a mix of various coding agents and assistants; different tools implied different repository structures, agent instructions, skills, and development workflows, making it difficult for teams to share knowledge and reuse capabilities. Cult.fit selected Codex as the unified foundation, incorporating repository structures, skills, testing, code reviews, and feedback loops into common standards. Kumar stated, 'Once standardization is complete, we can build the entire software development lifecycle around a single platform.'

The second stage involves extending capabilities beyond engineering. Cult.fit made its company data easier for Codex to process, enabling various teams to find answers and perform routine analyses themselves—previously, even small data requests had to go through a central analyst team, creating a bottleneck. The HR team built workflows using Codex to manage performance evaluations, scheduling, feedback collection, and summarization. Non-engineering teams built over 20 internal applications in recent months.

The third stage is organizational empowerment mechanisms. Kumar emphasized that enabling all employees to use AI cannot rely solely on issuing tool access permissions. Cult.fit’s approach was: first, host large-scale hackathons so employees could see the possibilities; then, organize one-day workshops for management, allowing leaders from various functions to build content relevant to their own work using Codex; the engineering team held office hours twice a week to help employees move ideas into production environments; and engineers were embedded in selected functional departments to provide technical and domain support.

Regarding cost savings, when the fees for internal dashboard software rose significantly, an engineer rebuilt the functionality using Codex over a weekend. Cult.fit now runs its own application, avoiding more than $100,000 in annual software costs. Kumar estimated that a social feed project previously took over a year to complete; with the new workflow, the team delivered it end-to-end in two months. In customer service, AI agents were integrated into the Sprinklr platform to handle requests such as waiving fines for missed classes, and the company reported a reduction in support costs of over 50%.

It should be noted that the above data all come from vendor case studies published by OpenAI, self-reported by Cult.fit. The 50% cost reduction does not disclose the calculation method, baseline composition, or whether it includes the costs of deploying Codex itself. The complexity and actual usage rates of the 20+ internal applications are also not explained. The comparison of shortening the social feed project from one year to two months is based on Kumar’s personal estimates rather than auditable project records. These are inherent boundaries of vendor case studies: they present only successes, not failures or total costs.

Fontes:OpenAI

Pesquisa

Google, Meta, and TikTok sue California AG to block teen social media law

Material
2026-11-06 05:10 GMT+8

Google (owned by Alphabet), Meta, and TikTok have filed separate lawsuits in US federal courts against California Attorney General Rob Bonta to block the enforcement of Assembly Bill 1709, a new state law designed to protect children under 16 from addictive social media features.

The law prohibits platforms from offering "psychologically exploitative features" such as personalized recommendations and autoplay to minors, and requires the deletion of children's accounts with those features. In their filings, the companies argued that the law violates the First Amendment's free speech protections and Section 230 of the Communications Decency Act. They contended that the statute wrongly compels them to adopt California's preferred methods of content presentation and substitutes the state's judgment for that of parents.

A Google spokesperson stated that overly broad laws break systems built to help kids find age-appropriate content. Meta argued that the state cannot override its editorial judgments about how to curate billions of posts. TikTok claimed that protected speech cannot be suppressed solely to shield young people from ideas deemed unsuitable by a legislative body.

The office of Attorney General Bonta said it was reviewing the lawsuits and remains committed to defending the law. Signed by Governor Gavin Newsom in September, AB 1709 is scheduled to take effect on January 1, 2027, unless a judge voids it or issues a stay. This legal challenge follows recent settlements where Meta agreed to pay up to $18 billion to resolve similar claims from nearly all US states.

Fontes:the Guardian

Pesquisa
Próxima →
Entrar / Cadastrar
LeituraPesquisaRadarFramework de investimento

Kumar describes the next step goal as an 'autopilot operations org'—where AI takes on more support work around frontline teams—and 'trainer in your pocket'—extending personalized guidance to a scale human coaches cannot cover. His core argument can be summarized as: software has been democratized, and organizations need supporting mechanisms to let this capability truly flow, rather than stopping at the issuance of tool licenses.