# Сопутствующие статьи по теме Claude

Новостной центр HTX предлагает последние статьи и углубленный анализ по "Claude", охватывающие рыночные тренды, новости проектов, развитие технологий и политику регулирования в криптоиндустрии.

Anthropic Cries Wolf: Is the AGI Threat Real, or Just an IPO Story?

Anthropic has published an article titled "When AI builds itself," discussing the emerging concept of "recursive self-improvement," where AI begins to actively participate in designing, training, testing, and optimizing its own subsequent versions. The company presents internal data showing that by May 2026, over 80% of code merged into its codebase was written by Claude, its AI model. Claude's capabilities have expanded to handling complex, open-ended engineering tasks, achieving a 76% success rate in such areas, and even contributing to research processes, such as optimizing code performance and conducting AI safety experiments. Anthropic outlines an evolution from human-driven development to AI-assisted workflows, culminating in the current stage where AI agents can autonomously write, run, and delegate code. The company cautions that the path toward a "closed loop," where AI continuously improves itself, is becoming visible. It calls for coordinated global mechanisms to potentially slow or pause frontier AI development to allow safety research and societal structures to catch up. However, the timing of this warning coincides with Anthropic's preparations for an IPO, framing the narrative not just as a safety concern but also as a demonstration of Claude's advanced capabilities and its integral role in accelerating Anthropic's own R&D—creating a potential "flywheel" effect for competitive advantage. This contrasts with OpenAI's recent, more policy-oriented discussion of the same risks, highlighting the competitive dynamics in the AI industry as companies position themselves in both the technological and regulatory landscape.

marsbit06/05 07:06

Anthropic Cries Wolf: Is the AGI Threat Real, or Just an IPO Story?

marsbit06/05 07:06

Breaking News! Anthropic Calls for a Universal Pause in AI Research

Anthropic warns of AI self-evolution, reporting that over 80% of its internal code is now written by its AI, Claude. Productivity has surged, with engineers merging 8x more code than in 2024. Claude's performance on complex, open-ended tasks jumped from 26% to 76% success in six months, nearing human parity. The company introduces a new metric: "AI task duration." In 2024, AI handled 4-minute tasks; by 2026, it manages 16-hour tasks, with capability doubling every 4 months. Claude also reviews code, catching bugs that previously caused outages, and significantly outperforms humans in research tasks like optimizing code (52x speedup) and conducting AI safety experiments. Anthropic outlines three potential futures: 1) Progress plateaus, 2) AI accelerates but humans remain in control, or 3) AI achieves full recursive self-improvement (RSI), designing its own successors. This final path could revolutionize fields like medicine but also risks catastrophic alignment failure if control is lost. The call echoes similar concerns from OpenAI. Anthropic proposes a coordinated pause on AI development—if a verifiable mechanism to ensure all labs comply can be established.

marsbit06/05 00:27

Breaking News! Anthropic Calls for a Universal Pause in AI Research

marsbit06/05 00:27

This Could Be the AI-Powered Siri We Get

Apple is set to unveil a major overhaul of Siri at its upcoming WWDC event, marking its most significant update since the AI assistant's debut in 2011. Faced with criticism for lagging behind competitors like ChatGPT and Google Gemini, the new Siri will feature a completely redesigned interface with a dark theme and chatbot-style interaction, deeply integrated with the Dynamic Island. Key upgrades include persistent conversation memory, addressing a long-standing user complaint. Most notably, Apple will reportedly allow third-party AI models, such as Google Gemini and Anthropic's Claude, to be integrated directly into Siri, transforming it into an AI model distribution platform. This strategic shift positions iOS not as having the single best AI model, but as the best platform for accessing and utilizing various AI models through superior system-level integration. Apple's approach leverages its strengths in hardware-software integration, privacy, and access to user data (contacts, calendar, photos) to create a differentiated experience, even while potentially relying on external infrastructure like Google's for some queries. This move represents a calculated bet that the ultimate AI advantage lies not in having the most powerful model, but in which system can integrate and utilize AI most seamlessly for the user. The success of this strategy will be tested by whether the new Siri can win back users who have grown accustomed to more advanced standalone AI tools.

marsbit05/29 07:15

This Could Be the AI-Powered Siri We Get

marsbit05/29 07:15

Claude Opus 4.8 Just Launched, Immediately Claimed to Be DeepSeek and Qwen

Anthropic has launched Claude Opus 4.8, an upgraded flagship AI model, while announcing a massive $65 billion Series H funding round, valuing the company at $965 billion. Claude Opus 4.8 offers improvements in coding, reasoning, agent tasks, and knowledge work, while maintaining previous pricing. Key enhancements include greater reliability in multi-step tasks and a significant increase in honesty—it's more likely to flag uncertainties or potential code issues. However, some user tests noted instances where the model incorrectly identified itself as other AIs like DeepSeek or Qwen, suggesting possible distillation behavior. Alongside the model, Anthropic introduced new features, most notably 'dynamic workflows' in Claude Code. This research preview allows the system to coordinate dozens to hundreds of parallel sub-agents for large-scale tasks like major code migrations, framework replacements, and security audits. The $65 billion funding round, which includes strategic investments from hyperscalers like Amazon and key hardware partners, is intended to secure massive compute capacity. This positions Anthropic to shift from a model provider to a company building a comprehensive enterprise AI infrastructure system, integrating intelligence, development tools, cloud platforms, and compute resources.

marsbit05/28 23:56

Claude Opus 4.8 Just Launched, Immediately Claimed to Be DeepSeek and Qwen

marsbit05/28 23:56

Just Now, Chinese AI Enters Top 2 in Global Programming, Only Claude Remains Ahead

**China's AI Ranks Second Globally in Programming, Trailing Only Claude** Today, Alibaba's Qwen3.7-Max achieved a score of 1541 on the Code Arena benchmark, securing fourth place globally and surpassing top models like GPT-5.5 and Gemini 3.5 Flash. Among the top positions, it is now the only non-Claude model, placing second overall after Anthropic's Opus models. Before this official ranking, Qwen3.7-Max had already gained recognition overseas. In practical tests, it outperformed rivals on tasks like creating a self-training Tetris AI and generating complex 3D models, often at a significantly lower cost. Developers praised its ability, especially when integrated with tools like Hermes Agent and OpenCode, to effectively replace models such as GPT-5.5. In a hands-on challenge to create a 3D racing game from a detailed prompt, Qwen3.7-Max delivered a fully playable HTML file in the first attempt, requiring only minor bug fixes. It uniquely included a start menu and sound effects—details missed by other models. While competitors like Gemini 3.5 Flash and Claude Opus 4.6 produced less polished or functional versions, and GPT-5.5 had its own quirks, Qwen3.7-Max stood out for its initial completeness and playability. This performance stems from its design as an "Agent Base Model," built for long-duration, autonomous task execution. Internal tests show it can run continuously for 35 hours, making over 1158 tool calls without context degradation or instruction drift. Key technical advancements include "environment expansion" training, which improves adaptability across different frameworks, and "long-horizon autonomous execution" training, enabling sustained strategic decision-making. By entering the top tier of the programming arena, Qwen3.7-Max demonstrates that Chinese AI models are not just catching up but are becoming defining competitors, challenging the long-standing dominance of Silicon Valley in this field.

marsbit05/27 00:17

Just Now, Chinese AI Enters Top 2 in Global Programming, Only Claude Remains Ahead

marsbit05/27 00:17

Microsoft to Discontinue Using Claude: Too Expensive, or a Calculated Move?

Microsoft is discontinuing internal access to the AI coding assistant Claude Code for its engineers, citing high costs and transitioning them to its own GitHub Copilot CLI. Reports indicate that Claude Code's usage-based pricing model led to significant expenses, with estimates of $500-$2000 per engineer monthly, prompting cost reviews even at large firms like Uber. While Microsoft officially attributes the move to cost, analysis suggests it may also be a strategic move. Microsoft allowed internal use of Claude Code for six months to benchmark it against Copilot, gathering feedback to improve its own product. This "experiment" highlights deeper challenges: Microsoft lacks a competitive in-house general-purpose AI model, heavily relying on external partners like OpenAI and Anthropic. Claude Code has gained substantial market traction, surpassing Copilot in features like context window size and task completion rates, and is eroding Microsoft's control over the developer ecosystem. The shift signals broader industry concerns over AI pricing models and the strategic risks of dependency on external, rapidly evolving AI tools.

marsbit05/25 10:28

Microsoft to Discontinue Using Claude: Too Expensive, or a Calculated Move?

marsbit05/25 10:28

Mythos Report Released: Billions of Devices Worldwide Exposed, 10,000 Critical Vulnerabilities Uncovered in 30 Days

The first report from Anthropic's "Project Glasswing" reveals staggering results from its secret initiative using the next-generation AI model, Claude Mythos Preview. In just 30 days, collaborating with roughly 50 global tech giants and critical infrastructure developers, Mythos identified over 10,000 high or critical-severity software vulnerabilities. It demonstrated an extremely low false-positive rate, even outperforming human experts, and successfully intercepted a $1.5 million bank fraud in progress. Key findings include uncovering 2,000 bugs in Cloudflare's core systems, fixing 271 critical vulnerabilities in Firefox 150 (ten times more than previous methods), and discovering a 27-year-old hidden bug in OpenBSD's codebase. The AI even autonomously constructed full attack chains for some exploits. Mythos also scanned over 1,000 essential open-source projects, identifying 23,019 total vulnerabilities, with 6,202 rated high/critical by the AI. Independent verification confirmed a 90.6% true-positive rate, validating 1,094 severe vulnerabilities. A critical case involved wolfSSL, a cryptography library used by billions of devices, where Mythos found a flaw allowing perfect digital certificate forgery. This unprecedented discovery speed has created a new crisis: human developers are overwhelmed and cannot patch vulnerabilities fast enough. In response, Anthropic is rolling out defensive tools like "Claude Security" to auto-generate patches and releasing frameworks to help security teams automate code review and threat modeling. Due to its immense power and potential for weaponization if misused, Anthropic is delaying Mythos's public release until robust safety measures are established. The company urges the industry to shorten patch cycles, enforce updates, and strengthen security fundamentals. The project signals a paradigm shift where AI could eventually make critical code vastly more secure, though the transition period poses significant challenges for human defenders.

marsbit05/25 00:09

Mythos Report Released: Billions of Devices Worldwide Exposed, 10,000 Critical Vulnerabilities Uncovered in 30 Days

marsbit05/25 00:09

The Revived Codex, Carrying OpenAI's Hopes for IPO

This article analyzes the intense recent development of OpenAI's Codex, positioning it as a crucial component for OpenAI's impending IPO. Over the past two months, Codex has seen a rapid series of major updates focused on integrating into real enterprise workflows. Key new features include enhanced context capture (Appshots, file previews, built-in browser), long-running task execution ("Goal Mode"), remote operation (phone control, lock-screen access), and enterprise management tools (plugin sharing, access tokens, automated risk review). These updates aim to make Codex a comprehensive AI workbench that can "see the scene, push tasks, and manage risks." The author argues that while ChatGPT proves OpenAI's massive user base and API provides foundational revenue, Codex represents OpenAI's clearest path to demonstrating tangible, high-value commercial viability. It targets developers and engineering teams—a segment already accustomed to paying for efficiency gains in costly software development cycles. This is critical because, despite higher overall revenue, OpenAI's adjusted operating margins remain deeply negative, highlighting the challenge of outrunning immense compute costs. The pressure is amplified by competitor Anthropic's success with Claude Code, which has shown that a focused approach on high-value enterprise and developer workflows can lead to a path toward profitability. Codex's aggressive evolution is thus seen as OpenAI's strategic move to capture a similar enterprise-ready, revenue-generating narrative essential for its market debut. In essence, "ChatGPT proved OpenAI has users. Codex needs to prove OpenAI is a business that can make money."

marsbit05/24 04:55

The Revived Codex, Carrying OpenAI's Hopes for IPO

marsbit05/24 04:55

How an Anthropic Engineer Saved 300 Million Tokens in a Week: A Claude Code Caching Guide

Anthropic engineers reveal how prompt caching in Claude Code dramatically reduces token consumption. By reusing already-processed context, cached tokens cost only 10% of regular input tokens. The author saved over 300 million tokens in a week, with 91 million cached tokens in a single day counted as roughly 9 million for billing. Caching works via prefix matching across three layers: system (instructions, tools), project (CLAUDE.md, rules), and conversation (chat history). As long as the beginning of a request matches cached content, Claude reuses it instead of reprocessing. Key points for users: - Claude Code's cache TTL is 1 hour (vs. 5 minutes for API/Sub-agents). Avoid pausing a session beyond this. - Switching models (including enabling "Opus plan" mode) breaks cache, forcing a full reprocess. - For task switching, use a clear session handoff instead of letting an old session expire. - Place large documents in Projects, not directly in chat, for better caching. High cache reuse benefits both users (longer sessions) and Anthropic (lower costs). Monitoring cache hit rates is crucial internally. By managing context as an asset and avoiding cache-breaking habits, users can make their Claude Code sessions more efficient and cost-effective.

marsbit05/24 00:36

How an Anthropic Engineer Saved 300 Million Tokens in a Week: A Claude Code Caching Guide

marsbit05/24 00:36

Anthropic Major Release: "The Founder's Playbook" - All 4 Stages of Entrepreneurship, Completely Reimagined with AI

**Anthropic Releases "The Founder's Playbook," Reimagining the Four Stages of Startups with AI** The logic of entrepreneurship is being fundamentally reshaped by AI. Anthropic's new handbook, "The Founder's Playbook: Building an AI-Native Startup," defines the AI-native startup as a new species: not a traditional company with AI tools, but a venture driven by AI from day one. The founder's role is transforming from a hands-on builder to a conductor or architect, orchestrating AI agents for execution while focusing on high-level judgment and strategy. Anthropic outlines a product matrix of Claude tools for different tasks: Claude Chat for interactive research, Claude Code for generating production-ready code, and Claude Cowork for automating knowledge-intensive workflows. The handbook structures the startup lifecycle into four stages, detailing core goals, pitfalls, and AI applications for each: 1. **Idea Stage**: Focuses on validating a real problem. The core challenge is avoiding confirmation bias. AI practices include using Claude as a "structured devil's advocate" to challenge assumptions and for automated market/competitor research. 2. **MVP Stage**: Aims to gather early signals of Product-Market Fit (PMF). Key risks are technical debt and scope creep due to rapid AI-assisted development. Recommended AI uses include maintaining project memory documents (e.g., CLAUDE.md), using Claude Code for structured coding, and automating user feedback analysis. 3. **Launch Stage**: Centers on establishing scalable growth, operations, and compliance. Challenges include accelerating technical debt and founders becoming bottlenecks. AI should be used to build an "operating system" for launch—automating routine tasks (scheduling, reporting, content) and code audits—freeing founders for critical decisions. 4. **Scale Stage**: Focuses on achieving sustainable business operations. The main challenge is delegating operational control. AI should be leveraged for differentiated marketing, operational optimization, and building competitive moats through data network effects. The handbook concludes that in the AI era, "Can we build it?" is no longer the primary constraint. The advantage shifts back to foundational strengths: **insight, judgment, and a deep understanding of a specific problem and audience.**

marsbit05/22 13:58

Anthropic Major Release: "The Founder's Playbook" - All 4 Stages of Entrepreneurship, Completely Reimagined with AI