Recent stories
Each story's summary, with the places it was discussed.
Summaries are written by AI from the linked sources. Open the source for full context.
Claude Fable 5.2 coordinated more than 1,000 AI agents to build a working operating system in eight days
Artificial Intelligence · Large Language Models · Anthropic · Claude · AI Agents
Claude Fable 5.2 coordinated more than 1,000 agents to build a bootable operating system in eight days, including a kernel, drivers, storage, graphical interface, installer, and Claude functionality.
It migrated a critical production service from Python to Rust, making startup approximately 26 times faster, achieved 64.9% on RoboDojo, and completed a computer-verifiable formalization of Fermat’s Last Theorem in 30 hours. Mythos 5.2 improved cybersecurity results but is restricted to vetted cybersecurity and biomedical organizations.
Claude Haiku 5.5
Artificial Intelligence · Large Language Models · Anthropic · Claude · LLM Benchmarks
Anthropic released Claude Haiku 5.5, its fastest, cheapest, and most capable small model, for high-volume tasks such as summarization, database queries, classification, browser use, and subagent coding.
It costs about 75% less to run than Haiku 4.5, supports adjustable effort, and scores 72.4% on OSWorld 2.1 and 39.2% on Terminal-Bench 4.0. It is available across major cloud platforms. Anthropic also halved Sonnet 5.5 cache-read prices, added monthly API credits for Max and Team subscribers, and introduced beta computer-use and browser-use support in its SDKs.
Anthropic’s open-source repository demonstrates how Claude Skills package instructions, scripts, and resources for specialized, repeatable tasks.
It includes creative, technical, enterprise, and document skills, alongside the Agent Skills specification and template. Skills can be installed through Claude Code plugins or used in Claude.ai and the Claude API, with guidance for creating custom skills.
Claude Fable 5.2 and Mythos 5.2
Artificial Intelligence · Large Language Models · Anthropic · Claude · LLM Benchmarks
Anthropic launched Claude Fable 5.2 and Claude Mythos 5.2, the same model with stronger cybersecurity and biology safeguards in Fable and vetted access to Mythos.
They improve coding, reasoning, vision, science, hardware, and robotics, including a 49.1% FrontierCode 2.0 score, 89.6% CVE Rediscovery, 64.9% RoboDojo, working electronic devices, black-hole results, and Fermat’s Last Theorem formalized in 30 hours. Outputs are up to 30% faster, cost $8/$40 per million tokens, and support 2.5× Fast mode.
Anthropic has released Claude Opus 5, a high-performance AI model offering superior coding, scientific research, and complex problem-solving capabilities.
It matches the cost of Opus 4.8 while providing significant performance gains and improved autonomous agent behavior. The model introduces enhanced safety guardrails and remains safer across most metrics, though it stays behind Mythos 5 in specific offensive cybersecurity tasks.
Anthropic has released Claude Opus 4.8, an upgraded AI model featuring improved reasoning, agentic reliability, and coding capabilities.
It introduces 'dynamic workflows' for large-scale tasks and new 'effort control' settings, allowing users to balance speed and intelligence. The update improves honesty and alignment, maintaining stable pricing while outperforming previous versions across diverse professional benchmarks.
Anthropic CEO Dario Amodei clarifies that the company does not advocate for banning open-weights models.
Instead, he proposes focusing on restricting chip access for authoritarian regimes, curbing industrial-scale model distillation, and implementing mandatory safety testing for all powerful AI models to address national security and safety risks.
After years using Claude Code, the author is replacing it with Pi because accumulated usability, stability, and product-direction frustrations outweigh its strengths.
Claude Code supports only one IDE-connected session, its VSCode GUI has missing features and cluttered subagent output, and frequent releases—244 in the year, over three-quarters changing prompts or tool descriptions—add overhead, including a nearly 3,000-token Bash description. Pi required initial setup but offers stability, minimalism, and extensive customization.
Redesigning Source Control at Anthropic
Backend · Distributed Systems · Software Architecture · Anthropic
Anthropic built Scion, an in-house source control system, and migrated its monorepo from git in five months as merged changes grew from 2,500 to over 30,000 per day.
Designed for always-online agents, Scion uses server-side changesets, content-addressed caching, API-based queries, and sparse or virtualized checkouts instead of fully local repositories. Its infrastructure uses 200 times less CPU and memory, makes CI and research checkouts two orders of magnitude faster, and cuts CI CPU use by 25%. A seven-person team built it with Claude.
Anthropic bans ‘abusive or cruel behavior’ toward Claude
Artificial Intelligence · Large Language Models · Ethics · Anthropic · Claude
Anthropic has updated its usage policy to address high-risk misuse and prohibit sustained, needless abusive or cruel behavior toward Claude, with conversation termination remaining the primary enforcement mechanism.
The rules also ban deceptive commercial or political campaigns, election misinformation and suppression, broader weapons software and autonomous arming, and nonconsensual surveillance or criminal-justice recommendations. AI connected to potentially dangerous physical hardware must be observable and stoppable by a qualified operator. Exceptions may apply to certain government contracts with safeguards.
Anthropic, a public benefit corporation focused on securing AI’s benefits and mitigating its risks, is launching an IPO of Class A common stock on Nasdaq under ticker “ANT.”
The roadshow starts November 10, the expected offering price is $70–$78 per share, and trading is expected to begin November 19, subject to regulatory clearance. Anthropic cites enterprise adoption through Claude, including 50,000+ Eli Lilly employees, 18x faster Salesforce API migration, and Walmart’s higher Sparky order values.
anthropics/claude-quickstarts
Artificial Intelligence · Large Language Models · Anthropic · Claude · AI Agents
Claude Quickstarts is a collection of customizable reference projects for building deployable applications with the Claude API and Claude Agent SDK.
It covers customer support, financial analysis, computer and browser use, autonomous coding, and numerous Claude Managed Agents integrations for Slack, Linear, Sentry, MCP, chat, scheduled briefs, knowledge wikis, and multiagent workflows. The repository distinguishes minimal toolset examples from fuller implementations, demonstrating streaming, tool calls, persistence, vaults, sandboxing, web/desktop automation, and production-oriented reliability patterns.
Anthropic trained Claude on illegally obtained Stanford professors' books
Artificial Intelligence · News · Law · Anthropic · Claude
A federal judge ruled that Anthropic illegally obtained and used books by Stanford professors and other authors to train Claude, while treating training on lawfully acquired books as fair use.
Anthropic’s $1.5 billion Bartz settlement covers nearly 500,000 works, but Tobias Wolff, Chang-Rae Lee and others opted out and sued for up to $150,000 per infringed work, potentially seeking an injunction. More than 70 AI copyright lawsuits continue, while Authors Guild advocates mandatory licensing and attribution.
Anthropic introduces Claude Sonnet 5.5, a faster, more efficient complement to Claude Opus 5.5.
It delivers major gains in agentic coding, knowledge work, computer use, and visual understanding, generating outputs 30%+ faster and costing up to 30% less per task. Enhanced cybersecurity safeguards, distillation protections, and broad cloud availability accompany the release.
Anthropic is implementing identity verification using Persona to ensure platform safety, compliance, and prevent abuse.
Users must provide valid government-issued photo IDs and a live selfie. Data is encrypted and managed according to strict privacy standards. Anthropic confirms that this verification data is not used for model training or shared with third parties.
anthropics/claude-plugins-official is Anthropic’s curated marketplace for high-quality Claude Code Plugins, organizing Anthropic-maintained plugins under `/plugins` and community or partner plugins under `/external_plugins`.
Plugins use a standard structure for metadata, MCP configuration, commands, agents, skills, and documentation, and can be installed through Claude Code. The repository also defines immutable plugin slugs, migration aliases, and `strict: false` skill-bundle entries for repositories exposing selected `SKILL.md` files.
Fuck Your AI Safeguards
Artificial Intelligence · Cybersecurity · Large Language Models · Anthropic · Claude
Claude Code blocked a senior product engineer from investigating and fixing a security issue in a service they control, citing AI safeguards.
The proposed Cyber Verification Program requires multiple business days, relinquishing data-retention exemptions, and government ID or passport submission, with no approval guarantee. The engineer argues these restrictions hinder defensive security work while criminals remain unrestricted, and reports that EU-hosted open-weight models, including GLM 5.3, identified, fixed, and verified the vulnerability instead.
Anthropic launched a life sciences research group and laboratory using Claude agents to mine DNA datasets, generate hypotheses, and guide human experiments.
In an early result, roughly 950 agents analyzed 210 million tokens and identified ART, a previously uncharacterized bacteriophage enzyme system combining a reverse transcriptase, partner gene, and CRISPR-like repeat array. Its function remains under investigation.
The curl project was analyzed using Anthropic's Mythos AI model to identify security vulnerabilities.
While Mythos identified five potential issues, only one was verified as a genuine vulnerability. The author concludes that while AI-driven analysis is significantly superior to traditional static tools, the hype surrounding Mythos exceeds its actual performance compared to existing AI solutions.
Anthropic's Claude AI blocks video game cloning and reverse-engineering — sweeping new restrictions also hit legitimate developers and basic troubleshooting tasks
Artificial Intelligence · Large Language Models · Anthropic · Claude
Anthropic's Claude Code and possibly Opus 5.5 are reportedly refusing legitimate game reverse-engineering and development tasks, potentially because of a new rule barring recreation of specific existing games.
The alleged policy allows original games, genres, and users' designs but not renamed clones. Claude Code's GitHub has received many complaints about refusals during development, mechanic replication, and patching, though reports vary and the rule's exact text lacks official confirmation.