Reading

Andrej Karpathy, a former OpenAI co-founder and current Anthropic researcher, announced on X that he built a website called HackerTube using Anthropic's Claude Opus 5.5 model. Motivated by dissatisfaction with YouTube's homepage recommendations, Karpathy stated that he filtered 6,454 repeatedly recommended videos from 20 years of Hacker News comments. He claimed the entire development process took approximately two hours.

According to reports by INSIDE and Startup Fortune, the tool processed around 43 million Hacker News comments to curate these videos, which are categorized into about 150 topics. In a blog post dated October 31, Karpathy explained the methodology: Opus 5.5 established the data extraction pipeline, while Sonnet handled filtering links into the final list. Karpathy performed basic tasks such as creating Google Cloud API keys, launching a VPS, and purchasing the domain, while delegating server updates, security hardening, deployment, and code optimization to Claude.

Designed to counter mainstream algorithmic recommendations, HackerTube requires no login, uses mid-video thumbnails instead of clickbait, and stores viewing history locally in the browser. Current information about the project relies primarily on Karpathy's personal statements and media relays, with no independent evaluation of its performance or user experience yet available.

Streaming giant Netflix is testing a new feature called "AI Discover" within its TV application, designed to help users find new shows and movies using artificial intelligence. The feature appears as a "Discover" tab in the top menu and is currently rolling out gradually to members who use Netflix in English on supported TVs in the United States.

Unlike traditional algorithmic suggestions, the first experiment under "AI Discover" surfaces content based on specific themes, such as "Something Light," "Family Movie Night," or "Quick Watches." When a user selects a theme, the system displays recommendations matching that category. If a user dislikes a recommendation, they can provide feedback, and the app will remember this preference to adjust future suggestions.

Netflix states that AI Discover is powered by "a new intelligence layer we’ve been building that responds to what members tell us in real time and keeps learning." This product update follows recent layoffs of approximately 1,000 employees at Netflix, amid slowing growth and reports of decreased user viewing time, highlighting the company's effort to boost engagement through technical innovation.

Japanese automotive parts supplier Aisin announced on Monday that it has developed a new semiconductor in collaboration with Tohoku University. The chip utilizes magnetic random-access memory (MRAM) technology and consumes approximately 98% less power than similar existing chips, according to the company. This breakthrough is expected to help process large amounts of data more efficiently, potentially improving AI capabilities within vehicles.

As reported by Nikkei Asia, Aisin stated that the chip achieves significant energy savings by combining magnetic technology with efficient memory architecture. MRAM is a type of non-volatile memory that retains data without continuous power supply, unlike traditional dynamic random-access memory (DRAM). It remains unclear when the chip will enter mass production or which specific vehicle models will incorporate it.

U.S. AI firm Anthropic disclosed financial results through September 30 in an updated S-1 registration statement filed on Monday. While revenue reached $17 billion, up from $11.5 billion in the previous quarter, operating losses widened primarily due to escalating compute costs.

A key metric shows that compute expenses consumed 58% of revenue in the third quarter, rising from 48% in the second quarter. This reversal in margin improvement occurs just ahead of the company's anticipated IPO, a period when investors typically expect scale efficiencies to boost profitability. The filing indicates that revenue generated per dollar spent on compute has fallen back to 1.72x from a peak of just over 2x earlier this year.

The primary driver is that demand growth outpaced the capacity secured under long-term agreements. To keep up, Anthropic relied heavily on 'neoclouds'—smaller infrastructure providers like Nscale, Akamai, and Volta that offer flexible but more expensive deals—as well as paying SpaceX $1.25 billion monthly for compute. These short-term commitments cost significantly more than comparable compute procured under multi-year arrangements.

Anthropic acknowledged in the filing that absolute compute and infrastructure expenses are expected to increase and remain its largest operating expense for the foreseeable future. Although the company is developing its own compute base under a project called 'Anthropic Direct,' high-cost short-term procurement remains necessary to support current demand levels.

The Gen Digital team published its development methodology for Norton Revamp in the AWS Machine Learning blog, with the most valuable aspect being its deliberate division of responsibilities between humans and AI. This product is an AI-driven reputation management platform for professionals, completed by a small engineering team within months and launched in March 2026.

The team clearly defined the division of labor: humans retain authority over architecture decisions, security reviews, business logic, and the design of personalization algorithms; AI handles code generation, documentation, test scaffolding, refactoring, and UI component creation. This is not vague AI assistance but rather distinct boundaries cut along specific work segments.

Their quantified results show a 3x increase in feature implementation speed, with boilerplate generation accelerated by 90%, documentation by 80%, test creation by 70%, and UI component generation by 65%. Using Claude, they produced more than 10 variants per design concept, compressing design-to-prototype time from weeks to hours, with 70% of UI code generated by AI. They tested five pricing models in the time previously required to test one.

However, the team noted the cost: reviewing AI output consumed 20%–30% of engineering time. This means the net speedup is not 3x, but rather the actual gain after deducting review costs.

The most critical insight is this: AI accelerated implementation, but humans designed how the AI Learning System should adapt to user goals. In other words, when the system needs to make decisions involving domain judgment and long-term effects—such as determining under what conditions to adjust recommendation strategies—AI does not replace the human design role. The team also specifically pointed out that product-side generative AI (Amazon Bedrock, Amazon Nova, AgentCore Web Search) and development-side generative AI serve different purposes, and they deliberately managed them separately.

The evidence level of this record is self-reported by the team: there is no external audit, no explanation of baseline measurement methods, and no complete context regarding team size or project duration. For engineering teams evaluating AI-assisted development, its value lies not in proving how much code AI can write, but in demonstrating where the irreplaceable boundary of human judgment falls in a specific project—not on abstract terms like 'architecture,' but on the adaptation logic of learning systems, which requires understanding user behavior over the long term to design.

On November 3, 2026, Maciej Szpunar, First Advocate General at the European Court of Justice, delivered his formal opinion in Case C-250/25 (Like Company v. Google). The opinion suggests that AI chatbots based on Large Language Models (LLMs), which use Retrieval-Augmented Generation (RAG) to search and summarize online news articles, may constitute acts of 'reproduction' and 'communication to the public' within the meaning of EU copyright law.

The case stems from a lawsuit by Like Company, a Hungarian operator of internet portals publishing news articles. The company alleged that AI services like Google Gemini infringed its exclusive right to authorize or prohibit the online reproduction and communication of its publications by generating summaries without permission or compensation. Szpunar's opinion addresses questions referred by a Hungarian court regarding whether such AI usage violates EU rules on press publisher rights.

While an Advocate General's opinion is not binding on the Court, it often significantly influences the final judgment. If the Court adopts this view, it could force AI companies to adhere more strictly to copyright licensing agreements in the European market, altering the economic relationship between AI providers and the news industry and increasing compliance costs and legal risks for AI products.

Evertrue Capital, founded by former LocalGlobe and Impact X Capital Partners investor Yvonne Bajela, has announced the closure of its debut fund (Fund I). The fund raised $17 million and was oversubscribed. It will focus on early-stage investments across the United Kingdom and Europe, with a primary emphasis on artificial intelligence (AI) and fintech sectors.

Bajela stated that the average check size would be approximately £350,000, with plans to support at least 20 companies over the next three years. She noted that the current AI boom has changed the economics of launching a company, making "the ability to build" more accessible, while the capacity to create enduring businesses remains rare. This insight drove her decision to launch an independent fund to act on her own convictions.

The fund secured backing from five institutional investors, including the British Business Bank. As part of the British Business Bank’s £400 million Investor Pathways initiative, £90 million was allocated to 10 first-time fund managers, including Evertrue. Approximately 80% of the limited partners (LPs) are based in the UK, with the remainder scattered across Europe, Asia, and the US.

While the US venture market is described as a "barbell" where capital concentrates among top-tier firms, PitchBook data shows that emerging fund managers in the UK raised $7.4 billion through the end of September, surpassing the $6.2 billion raised in all of 2025. Bajela believes this environment offers opportunities to invest early in the next generation of top British talent, particularly founders with unique problem insights and strong execution abilities.

When building enterprise-grade AI agent analytics, a recurring problem is that agents can write SQL but do not know the business meaning behind column names or what they are authorized to read. Common solutions involve deploying a separate semantic layer—such as a knowledge graph, data virtualization platform, or dimensional model—and having the agent query that layer. On November 3, 2026, the Amazon Web Services Machine Learning blog presented a different approach: placing the semantic layer directly inside the operational Postgres database itself, using native database COMMENT, GRANT, and row-level security features to define meaning and control access, eliminating the need for a second system.

The specific mechanism operates in three layers. The first layer is meaning: the AIDB extension’s Semantic Knowledge Base (SemKB) embeds the definition of every table, view, and column, along with COMMENT ON text, into a vector index, allowing agents to find the correct tables and columns via natural language search. The second layer is relationships: Postgres foreign keys are automatically crawled by SemKB into join paths, so agents do not need to guess associations based on column names. The third layer is governance: the AIDB Model Context Protocol (MCP) endpoint has no permission system of its own; it passes the Postgres username and password for each request as HTTP Basic authentication, and the database’s GRANT and Row-Level Security (RLS) policies determine what that connection can see.

The blog illustrates the role of the meaning layer with a concrete example. In the C360 dataset, the purchases.price column records the order total, not the unit price. Without a COMMENT ON statement clarifying this, the agent treats it as a unit price multiplied by quantity, calculating approximately $502,529 for the top customer, Carolyn Inworth—a wrong figure. Adding a comment stating 'Total amount the customer paid for the entire order, in US dollars. Not the unit price.' changes the answer to $27,991.66. The model did not change, the prompt was not altered, and the tools remained the same; only the indexed comment within SemKB changed.

Governance verification is more direct. The blog defines two Postgres login roles: mcp_agent_full can see all columns, while mcp_agent_restricted cannot see customers.ssn. Running the exact same prompt to list five customers' Social Security Numbers yields different results. The full role returns the SSN values for all five customers; the restricted role’s tool call is denied at the Postgres layer, returning 'permission denied for column ssn,' after which the agent informs the user that SSNs cannot be read. The prompt, tool catalog, SemKB, and agent code are identical; the only difference is the Basic authentication credentials injected by AgentCore Identity.

Architecturally, the agent runs on the Amazon Bedrock AgentCore runtime, using Claude for planning and calling MCP tools through the AgentCore Gateway. The Gateway forwards requests to an Application Load Balancer within the EDB Postgres AI Virtual Private Cloud (VPC), reaching the AIDB edb-endpoints MCP server. No Lambda functions or translation layers are required—AIDB natively speaks the streamable HTTP MCP protocol. Auditing also remains within the same database: AIDB writes OpenTelemetry traces for every SQL execution into the aidb_otel schema, which can be queried using standard SQL.

On November 3, 2026, OpenAI announced the launch of a program called "Grid Protectors," aimed at reducing the power consumption of its data centers during peak hours to ensure local grids are not overwhelmed. According to Tom's Hardware, this move is intended not only to safeguard residential electricity but also to allow its data centers to connect to the grid and go online sooner, without waiting for additional power sources to come online.

OpenAI is testing the system at its planned campus in Effingham County, Georgia. The company stated that it has already signed a 3.2-gigawatt agreement with Georgia Power, which includes 1,000 megawatts of flexible demand response. This means that on the hottest days of the year, if electricity demand surges, this 1-gigawatt buffer would be enough to power approximately 800,000 households, equivalent to nearly 20% of the state's homes.

This technology, known as demand response, has been around for decades, typically involving grid operators manually calling up large industrial customers like steel mills to cut consumption. Today, with modern technologies and IoT, data centers can automatically reduce their load in under a minute. Nvidia has previously experimented with a similar system in the UK and proven its effectiveness. OpenAI stated that it intends to make this kind of flexibility a core part of how it builds and operates future campuses, working with utilities to agree on how much demand can be reduced, how quickly, and for how long.

US tech giant Google announced on November 3, 2026, that it plans to default-disable AI tools for K-12 students next year unless school administrators explicitly turn them on. The change affects Gemini, NotebookLM, Gemini in Classroom, and Google Search.

The critical constraint is that because AI features are deeply integrated into the search engine, schools choosing to restrict AI altogether will no longer be able to access Google Search. Chris Phillips, Vice President for Education at Google, told The New York Times that "Search and AI are deeply integrated," meaning if a school chooses to restrict AI, students "will not have access to search, and they’ll see a message from their school that indicates it’s not available."

This creates a dilemma for schools. While teachers may not want students using AI Overviews in Google Search to conduct research or complete assignments, they likely still want them to have access to the search engine to find sources and other learning materials. However, Google Search and its AI features are now inseparable. Given the dominance of Google’s suite of apps in classrooms and the widespread use of Chromebooks, the decision underscores the company’s influence over the education sector. Schools that want to continue having access to Google Search will have to accept its AI features.

"We’ll share more specifics with schools in the coming months ahead of the changes rolling out so they have time to prepare and enable a smooth transition," Phillips wrote in a blog post. "Our goal is simple: provide school leaders the control and confidence to use AI products in classrooms as they see fit."

Google also launched Guided Learning alongside this announcement. Unveiled earlier this year, the tool functions like an AI tutor designed to help users build deep understanding rather than just getting answers, responding to concerns that AI chatbots undermine the learning process by providing direct answers.

Mirae Asset Securities announced on October 1, 2026, that it is officially renaming its mobile trading system (MTS), previously known as 'M-STOCK', to 'MAPS'. The platform has been upgraded to a comprehensive investment hub centered on portfolio management. This move aims to leverage artificial intelligence to optimize users' decision-making processes, shifting focus from traditional quote viewing and order execution to smarter asset allocation support.

The new MAPS platform highlights two core sections on its home screen: 'Discovery' and 'My Investment'. The 'Discovery' section introduces an AI service called 'Market Now', which automatically generates a one-sentence summary of market trends every two hours, helping users quickly capture key information. Additionally, the platform has added a digital assets menu providing relevant market data and professional research materials, while expanding coverage to include information on global innovative companies such as US non-listed stocks.

Mirae Asset Securities stated that the long-term goal of MAPS is to realize an 'AI Everywhere' environment, ultimately evolving into 'Agentic WM' (Agentic Wealth Management). In this model, AI will not only provide information aggregation but also proactively monitor market changes and suggest investment actions based on user goals and constraints. Currently, the company's internal Tech & AI division, comprising about 40 personnel, is advancing this strategy, with over 85% of this year's targets for internal efficiency and customer service reportedly achieved.

According to WIRED, independent research firm Forum AI found that nearly 40% of answers provided by mainstream AI chatbots—including OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini—contained at least one factual error when responding to questions about the US midterm elections. For instance, Gemini incorrectly stated the mail-in ballot deadline for Pennsylvania, and Claude confused North Carolina Congressman Don Davis with a former school superintendent of the same name.

Research by NewsGuard further indicated that these models frequently cite “pink slime” websites—low-quality sources funded by political or ideological groups but presenting as mainstream news—when asked about candidates and policies. ChatGPT cited such sites in more than 70% of cases, while xAI’s Grok performed better with a 30% rate.

Pew Research data shows that approximately 8% of US adults used AI tools to find information on candidates and policies ahead of the midterms. Despite OpenAI publishing election safety guidelines, the findings highlight significant reliability risks in current AI technology when handling real-time, complex political facts, potentially misleading voter decision-making.