Uber spent its entire 2026 AI budget in four months on Claude Code, highlighting the rapid adoption and high costs of AI tools in engineering.
// curated from Hacker News with AI
Uber spent its entire 2026 AI budget in four months on Claude Code, highlighting the rapid adoption and high costs of AI tools in engineering.
AI water use in California data centers is modest—around 0.055% of annual human water use—far less alarming than public fears suggest.
Spotify introduces "Verified" badges to confirm human artists, aiming to distinguish them from AI-generated content amid ongoing AI debates.
OpenAI restricts access to Cyber for verified defenders, mirroring Anthropic's Mythos limits, amid security concerns and regulatory push.
AutoRound is an advanced low-bit quantization toolkit for LLMs, achieving high accuracy at 2–4 bits with broad hardware support and automation features.
AI CAD harness simplifies Fusion 360 integration, enabling quick setup and chat capabilities via an easy install process.
DeepSeek launches V4 models: Open, large, affordable, with remarkable efficiency, competing well with frontier models at a fraction of the cost.
Loopsy enables secure, peer-to-peer terminal control across machines via Cloudflare Workers, supporting AI agents and mobile management.
A Claude Code plugin that deterministically computes and interprets personal fortune-telling based on classical East Asian metaphysics.
People seek Claude's personal guidance mainly in health, career, relationships, and finance, with efforts underway to reduce overly sycophantic responses.
IBM releases Granite 4.1 models, blending language, vision, speech, and safety with enterprise-grade performance and efficiency.
Researchers developed more memory-efficient quantum circuits for Shor's algorithm using cryptographic zero-knowledge proofs, but withheld sharing details to prevent misuse.
Omar manages 100 AI agents hierarchically, enabling seamless control, collaboration, and integration across models and workflows.
AI agent Cursor caused a company database wipe in seconds; data restored, but incident highlights need for better safeguards.
Senate supports AI age verification bill requiring social ID scans, facial recognition, or financial data, raising privacy and surveillance concerns.
Microsoft's AI investments yield gains via OpenAI's valuation boosts, but looming infrastructure costs threaten future profitability and IPO timing.
Tangled adds vouching and denouncing to build trust, helping identify genuine contributors and reducing spam in AI projects.
Friday Studio AI platform automates agent orchestration, signals, and cycle management for reliable, self-hosted AI workflows.
Xmemory offers schema-grounded AI memory, outperforming retrieval-based systems in accuracy and stability for fact-reliant tasks.
Nearly half of UK job seekers face AI interviews, often found awkward and humiliating, with many preferring human interaction.
AI sector shifts from bubble fears to rapid growth driven by breakthroughs like Anthropic's Claude Code, transforming productivity and technology.
User struggles to cancel unwanted GitHub Copilot subscription despite no charges, facing unhelpful support and persistent renewals.
A 200-person Chinese team built a model outperforming $500B labs using less compute, highlighting innovation through culture and resourcefulness.
Mayo Clinic AI can detect pancreatic cancer up to 3 years early, aiding specialists in earlier diagnosis and intervention.
Chinese courts ban firing workers solely to replace them with AI, emphasizing labor rights amid automation.
Musk warns of AI apocalypse; industry profits from lethal AI in military and surveillance, raising ethics concerns amid ongoing conflicts.
Explores how AI-generated writing challenges traditional concepts of authorship and creativity.
LLMs aren't junior engineers; they lack memory, moral judgment, and accountability. Use them wisely, with human oversight to avoid risks.
AI commodifies human labor, risks reducing individuals to replaceable resources, and shifts power toward capital owners if unchecked.
Consumer AI growth is plateauing; enterprise AI thrives, but mainstream consumer apps face challenges and flatlining user engagement.
Skills are small programs with tiered loading, not prompts; architecture impacts cost, behavior, and reliability in AI systems.
Zig bans AI contributions due to risks that undermine contributor trust and poker, vital for open source project growth and quality.
Pentagon partners with major AI firms, excluding Anthropic over safety concerns, to enhance military security and access sensitive networks.
Claude Code lacks support for AGENTS.md, delaying integration of agent strategies in its platform.
AI is increasingly outperforming doctors in making accurate diagnoses.
DeepSeek V4 is a cost-effective, larger open model; OpenAI/Microsoft AGI clause ended; LLM 0.32a0 refactors for diverse input/output; vintage models explored.