Programmers will document for Claude, but not for each other
Hacker News (AI keywords)9 days agoOpinionThe post responds to complaints that programmers now write detailed CLAUDE.md and PROJECT.md files for AI, but not for coworkers. The author describes using Claude to maintain handoff notes between sessions and generate final high-level project summaries. His advice is to review those documents carefully, then commit them to the repository because they may help future maintainers.
Anthropic Co-founder Ben Mann Visits Taiwan to Discuss AI Safety and Claude Strategy
INSIDE 硬塞 AI9 days agoBusinessAnthropic co-founder and Anthropic Labs lead Ben Mann made his first visit to Taiwan, according to INSIDE. The report highlights his role in leading Claude Code and the Model Context Protocol, two key parts of Anthropic’s developer-focused product direction. The discussion centered on Claude strategy, AI safety boundaries, jobs, and Taiwan’s strategic role in the AI landscape.
Trend Micro Joins Anthropic Project Glasswing to Defend Taiwan’s AI Supply Chain★ 72
INSIDE 硬塞 AI9 days agoBusinessAnthropic introduced Project Glasswing after Claude Mythos Preview showed the ability to rapidly find high-risk vulnerabilities and generate connected attack commands. Trend Micro’s TrendAI has joined the framework, becoming the first Taiwanese cybersecurity vendor to do so. The article frames the move around Taiwan’s strategic AI hardware role and a new defensive logic: using AI to counter malicious AI.
Fine-tuning an LLM to write docs like it's 1995
Hacker News (AI keywords)9 days agoTutorialThe author builds a corpus from old Microsoft manuals, cleans OCR text, generates instruction-style JSONL examples, and fine-tunes Llama 3.1 8B and Qwen 2.5 7B with QLoRA. Tests cover malloc(), a fictional Win32 API, and a deliberately anachronistic REST API prompt. Qwen fine-tunes transfer the period documentation style best, but the experiment also shows hallucination risks, tuning complexity, and why these models augment rather than replace technical writers.
Microsoft AI Chief Says Anthropic Is Too Expensive as Enterprise AI Costs Rise
INSIDE 硬塞 AI9 days agoBusinessMicrosoft AI chief Mustafa Suleyman reportedly criticized Anthropic’s models as unacceptably expensive, highlighting rising enterprise AI costs. The article frames this as part of a broader “AI tax” problem, with companies reassessing ROI as vendor pricing pressure grows. Microsoft’s MAI models are presented as a potential internal alternative to reduce reliance on costly external providers.
Ahead of its IPO, Anthropic’s Daniela Amodei shrugs off doubts about AI returns★ 76
TechCrunch AI10 days agoBusinessTechCrunch reports that Anthropic has confidentially filed for an IPO while private investor demand remains strong. Co-founder Daniela Amodei said frontier AI companies need large amounts of capital because model training and inference are expensive. She also downplayed doubts about enterprise AI returns, arguing businesses are still early in learning how to use AI effectively, and explained why Anthropic prefers not to overbuild its own compute infrastructure.
Show HN: Formally verified polygon intersection, Opus 4.8 one-shot
Hacker News (AI keywords)10 days agoNew ToolThis GitHub project presents a formally verified multipolygon intersection algorithm checked in Lean 4. The author argues trust comes from the Lean checker and a small human-reviewed specification, not from trusting LLM output directly. It also documents how Claude Opus versions improved on Lean proof work, with Opus 4.8 reportedly completing larger proof strategies that earlier attempts could not.
Reality: The Final Eval — Lukas Petersson and Axel Backlund of Andon Labs
Latent Space10 days agoBenchmarkLatent Space talks with Lukas Petersson and Axel Backlund of Andon Labs, the authors behind VendingBench. The episode focuses on evaluating Claude models across a range from Haiku to Mythos. It also discusses how they build frontier evals from scratch, with an emphasis on creating benchmarks that remain useful and meaningful over time.
Show HN: Boxes.dev: ditch localhost; run Claude Code and Codex in the cloud
Hacker News (AI keywords)10 days agoNew ToolBoxes.dev appeared on Hacker News as a Show HN post, positioning itself as a way to move Claude Code and Codex workflows from localhost to the cloud. Based only on the title, it seems aimed at cloud development or remote agent execution. The provided source does not include details on architecture, pricing, security, integrations, or limitations.
My Agent Skill for Test-Driven Development
Hacker News (AI keywords)10 days agoTutorialJason Swett argues that uncoached AI agents still tend to write poor tests: vague, overcomplicated, tautological, or performative. His personal TDD skill guides agents through a specify-encode-fulfill loop inspired by Kent Beck’s Canon TDD. He also uses separate test and software design review skills, sometimes with Claude, to catch weak test design and prompt cleanup before implementation.
Reve 2 and Ideogram 4: Layouts in Imagegen
Latent Space10 days agoReleaseLatent Space’s roundup frames image composition as a major barrier now being tackled by layout-aware image models. Reve 2.0 emphasizes precise generation and editing with layouts, while Ideogram 4.0 uses bounding boxes tied to region descriptions. The issue also covers MAI-Thinking-1, Gemma 4 12B, open audio models, agent execution layers, and model-routing cost debates.
I built a vulnerable app and spent $1,500 seeing if LLMs could hack it
Hacker News (AI keywords)10 days agoBenchmarkThe author built a vulnerable React Native app with a Python backend and a Firebase access-control flaw. GPT 5.5 solved 7 of 10 runs, while Deepseek and Claude variants solved fewer attempts. Many other models failed due to refusals, API-focused tunnel vision, false positives, or inability to use the exposed Firebase path correctly.
How we contain Claude across products★ 74
Hacker News (AI keywords)10 days agoCommentaryAnthropic describes containment as the core security strategy for increasingly capable Claude agents. The post compares ephemeral containers for claude.ai, OS-level sandboxing and approvals for Claude Code, and VM isolation for Claude Cowork. It also details missed risks, including pre-trust project config execution, user-delivered prompt injection, exfiltration through approved domains, and reduced enterprise visibility inside VMs.
Lovable signs multi-year deal with Google Cloud to up usage 5x, source says
TechCrunch AI11 days agoBusinessTechCrunch AI reports that Lovable and Google signed an expanded multi-year agreement. The deal reportedly includes a fivefold expansion of Lovable’s footprint on Google Cloud. It also includes expanded access to Anthropic Claude, though the article does not specify contract value, timing, exact Claude usage, or any immediate product changes for users.
How LLMs Actually Work
Hacker News (AI keywords)11 days agoTutorialThe article explains how modern LLMs convert text into token IDs, embeddings, and position-aware vectors before passing them through stacked transformer blocks. It covers attention, multi-head attention, KV cache, GQA, feed-forward networks, MoE, residual streams, normalization, and decoding. Its goal is educational: helping readers understand the common architecture behind many current model families and read model cards or papers more confidently.
No, Artificial Intelligence Is Not Conscious★ 72
Hacker News (AI keywords)11 days agoOpinionTed Chiang criticizes the anthropomorphic framing around Anthropic’s Claude and its constitution. He argues that LLMs are sentence-continuation systems producing fictional conversational roles, not entities with subjective experience. The essay warns that presenting chatbots as morally aware risks misleading users and shifting responsibility away from humans and companies.
Uber Caps Usage of AI Tools Like Claude Code to Manage Costs
Simon Willison's Weblog11 days agoBusinessUber has reportedly capped employee token spending at $1,500 per month for each agentic AI coding tool, including Cursor and Claude Code. Simon Willison frames this as a rational response to overspending, especially after earlier discussion that Uber exhausted its 2026 AI budget in four months. He estimates that two actively used tools would imply a $36,000 annual cap per engineer, about 11% of median US Uber software engineer compensation.
Microsoft Build: MAI-Thinking-1 and MAI Family Models★ 78
Latent Space11 days agoReleaseMicrosoft used Build to present itself as both an AI platform and a first-party model lab, announcing seven MAI models across reasoning, code, image, transcription, and voice. The standout was MAI-Thinking-1, described as a 35B active MoE with 256K context and clean data lineage. The recap also ties the launches to GitHub Copilot, Windows agent runtime ambitions, Web IQ grounding APIs, Foundry distribution, and MAIA 200 hardware.
Claude Code Lead Boris Cherny Says His Code Is 100% AI-Written
INSIDE 硬塞 AI11 days agoCommentaryClaude Code lead Boris Cherny says his code is now 100% written by AI while he runs hundreds of agents in parallel. The article frames engineers less as manual coders and more as conductors who define problems, review outputs, and shape architecture. It highlights a broader shift in software development workflows driven by AI coding agents, without presenting detailed benchmarks or implementation data.
Show HN: Paseo - Beautiful open-source coding agent interface
Hacker News (AI keywords)12 days agoNew ToolPaseo provides one interface for tools such as Claude Code, Codex, Copilot, OpenCode, and Pi. It runs agents through a local daemon on the user's own machine and supports desktop, mobile, web, and CLI clients. Its appeal is multi-agent orchestration and cross-device control, though real adoption depends on workflow fit, security, and reliability.
Microsoft's new MAI models★ 72
Simon Willison's Weblog12 days agoReleaseMicrosoft announced MAI-Thinking-1, a 35B reasoning model available to select early partners, and MAI-Code-1-Flash, a 5B coding model rolling out to GitHub Copilot individual users in VS Code. Simon Willison highlights their relatively small parameter counts and Microsoft's claim that MAI-Thinking-1 was preferred to Sonnet 4.6 in internal blind evaluations. He also questions what Microsoft's clean and appropriately licensed training data claims mean in practice.
Microsoft announces Scout, an autonomous AI agent built on OpenClaw
Hacker News (AI keywords)12 days agoNew ToolMicrosoft unveiled Scout at Build as a new “autopilot” agent for Microsoft 365. It can connect across Teams, Outlook, OneDrive, and SharePoint, use an Entra identity, and interact with external apps through MCP. The release is experimental for Frontier customers, with security controls required. Analysts warn Scout may amplify existing governance problems because it can act on data, not merely surface it.
Anthropic scales Claude Mythos to critical infrastructure in 15+ countries★ 72
TechCrunch AI12 days agoBusinessAnthropic is expanding its Project Glasswing security vulnerability program and access to Mythos. The rollout covers 150 organizations across 15 countries, focusing on power, water, healthcare, and communications infrastructure. The company is targeting sectors where a cyberattack could affect as many as 100 million people, although implementation details and participating organizations were not disclosed in the provided text.
Please don't spam people looking for employment. It's just cruel
Hacker News (AI keywords)12 days agoEthicsA Hacker News poster says they received a self-promotional AI/LLM services email shortly after posting in a job-seeking thread. The email appeared to exploit the context of their search, turning a moment of hope into another discouraging spam interaction. The discussion broadened into concerns about AI-generated cold outreach, recruiter spam, cybersecurity pitches, and the need for basic empathy in automation.
Expanding Project Glasswing★ 76
Hacker News (AI keywords)12 days agoBusinessAnthropic is expanding Project Glasswing, its program for using Claude Mythos Preview to find vulnerabilities in critical software. The new cohort includes around 150 organizations across more than 15 countries, including infrastructure providers, vendors, nonprofits, and open-source maintainers. Anthropic frames the expansion as preparation for a world where powerful cyber-capable AI models become cheaper and more widely available, shifting focus from finding bugs to validating, disclosing, patching, and deploying fixes.
Michael Burry says neither SpaceX nor Anthropic is worth $1T
Hacker News (AI keywords)12 days agoBusinessBased only on the headline, Michael Burry argues that neither SpaceX nor Anthropic is worth $1 trillion. The item appears to sit at the intersection of private-market valuations, AI enthusiasm, and skepticism toward highly priced technology companies. Without the article text, the specific reasoning, valuation framework, or any detailed comments about Claude or an AI bubble cannot be verified.
Pasted File Editor
Simon Willison's Weblog12 days agoNew ToolSimon Willison released Pasted File Editor, a browser prototype inspired by Claude's handling of large pasted text. Instead of filling the editor with a large paste, the tool turns the content into a file attachment. It also supports opening files directly, dragging files onto the interface, and displaying images as thumbnails. Codex desktop helped build the prototype.
AI Agent Guidelines for CS336 at Stanford
Hacker News (AI keywords)13 days agoEthicsStanford CS336’s CLAUDE.md sets boundaries for AI coding assistants such as ChatGPT, Claude Code, GitHub Copilot, and Cursor. Agents may explain concepts, review student-written code, suggest debugging checks, and point to course materials. They should not write code, complete TODOs, edit repositories, run shell commands, or implement core assignment components for students.
Ask HN: Who wants to be hired? (June 2026)
Hacker News (AI keywords)13 days agoBusinessThis is Hacker News’ June 2026 “Who wants to be hired?” thread for individuals actively looking for work. Posters are asked to share location, remote preference, relocation willingness, technologies, resume or CV, and email. Visible comments include developers, full-stack engineers, data science consultants, systems engineers, and designers, with some mentioning LLM integration, RAG, AI agents, Gemini API, and Claude tool calling as part of their experience.
Launch HN: Expanse (YC P26) - Unlock Wasted GPU Capacity
Hacker News (AI keywords)13 days agoNew ToolExpanse is a YC P26 launch for improving effective utilization in SLURM and Kubernetes GPU/HPC clusters. It analyzes source code, job scripts, hardware topology, and telemetry before submission to recommend GPU VRAM, CPU, memory, utilization, and walltime. The team says it also detects likely failures, offers line-level optimization hints, and fine-tunes cluster-specific models over time.