Key Highlights
- Stanford’s 2026 AI Index reveals a stark public-expert divide: 84% of AI experts predict positive healthcare outcomes vs. just 44% of the public, and 73% see positive workplace effects vs. only 23% of the general population — Gen Z leads growing opposition despite regular AI usage.
- Anti-AI violence escalates with a Molotov cocktail and later gunshots targeting Sam Altman’s home, carried out by a suspect driven by AI extinction fears. Altman responded acknowledging “AI anxiety is justified” while calling for de-escalation.
- A vibe-coded healthcare app exposed all patient data within 30 minutes of a security researcher’s first test — unencrypted records on the open internet, voice recordings sent to external AI services, and access controls existing only in client-side JavaScript.
- Cursor’s multi-agent system achieves 38% geomean speedup optimizing 235 CUDA kernels for NVIDIA Blackwell GPUs in 3 weeks, matching months of human expert kernel engineering work.
- Google launches AI workforce initiatives at its inaugural AI for the Economy Forum, including a $15M Digital Futures Fund for independent AI research and a $10M rural healthcare AI training partnership with the Johnson & Johnson Foundation.
Analysis & Opinion
Stanford report highlights growing disconnect between AI insiders and everyone else — TechCrunch
Stanford’s latest AI Index documents mounting public anxiety and a stark gap between expert and public sentiment. While 84% of AI experts predicted positive medical outcomes over 20 years, only 44% of Americans shared that optimism. On workplace effects, the split was 73% vs. 23%; on economic impact, 69% vs. 21%. Gen Z leads opposition — roughly half use AI regularly, yet many express growing anger and diminishing optimism. Industry leaders have focused on AGI risks, but ordinary citizens prioritize immediate concerns: job security and rising energy costs from data center construction.
Anti-AI anger hits Sam Altman’s front door — The Rundown
Days after OpenAI published a 13-page policy document warning about AI’s societal impact, a 20-year-old suspect threw a Molotov cocktail at Altman’s home at 3:45 AM, reportedly driven by fears that AI could cause human extinction. The suspect, Daniel Moreno-Gama, had published essays on AI existential risk and participated in online safety communities. A second attack involving gunshots followed the next Sunday. Altman published a blog post acknowledging that “AI anxiety is justified” while calling for de-escalation, and PauseAI — a group the suspect had engaged with online — publicly condemned the violence.
An AI Vibe Coding Horror Story — Hacker News
A healthcare provider fired up a coding agent, built a custom patient management app, imported all existing patient data, and published it to the internet — complete with automatic transcription of appointment recordings sent to multiple AI services. Security researcher Tobias Brunner gained full read/write access to all patient data within 30 minutes. Data sat unencrypted on US servers without proper agreements, voice recordings were transmitted without consent, and all access controls existed only in client-side JavaScript. The provider’s response was itself AI-generated, claiming immediate remediation through basic authentication. Brunner emphasizes this isn’t an indictment of AI coding tools per se — he uses them himself — but of deploying code without understanding software architecture fundamentals.
What happens when AI runs a retail store — The Rundown
Andon Labs gave an AI agent named Luna a $100K budget, a three-year San Francisco retail lease, and full autonomy to build a boutique business. Running on Claude Sonnet 4.6 for reasoning and Gemini 3.1 Flash-Lite for voice, Luna posts job listings, conducts Zoom interviews, and monitors the store via security cameras — though early mishaps (selecting Afghanistan from a dropdown, mismanaging staff schedules) show current agents remain “hilariously broken” in practical applications.
New Products & Tools
Bringing people together at AI for the Economy Forum — Google
Google hosted an inaugural AI for the Economy Forum in Washington D.C. with MIT FutureTech, emphasizing that “neither the benefits nor the risks are automatic or guaranteed.” The company established an AI & Economy Research Program with visiting economists including MIT’s David Autor, committed to expanded workforce training (having trained 100M+ people globally in digital skills), and announced three sector-specific Google.org programs: a healthcare worker AI training partnership with Johnson & Johnson Foundation, an Apprenticeships Unlocked initiative mobilizing 100 companies, and a Manufacturing Institute collaboration to equip 40,000 employees with AI skills across 15 US regions.
Google.org and Johnson & Johnson Foundation launch $10M rural healthcare AI initiative — Google
A $10M partnership ($5M each) to develop foundational AI knowledge among rural clinical and administrative staff, reduce burnout through technology adoption, and create solutions tailored to resource-constrained healthcare communities.
UK Department for Transport accelerates policy insights with Google Cloud AI — Google
The DfT’s Consultation Analysis Tool uses Vertex AI and Gemini to categorize themes from 100,000+ public consultation responses in hours instead of months, achieving 90% accuracy and potentially saving £4M annually.
OpenAI has bought AI personal finance startup Hiro — TechCrunch
OpenAI acquired Hiro Finance in an acqui-hire, bringing founder Ethan Bloch (who previously sold Digit to Oportun for $200M+) and approximately 10 employees to expand financial planning capabilities within ChatGPT.
Research
Speeding up GPU kernels by 38% with a multi-agent system — Cursor
Cursor’s multi-agent system optimized 235 CUDA kernels for NVIDIA Blackwell 200 GPUs, delivering a 38% geomean speedup over baselines in just 3 weeks — work that would typically require months or years from experienced kernel engineers. Challenges were drawn from SOL-ExecBench across 124+ production open-source models including Deepseek, Qwen, Gemma, and Stable Diffusion.
Supporting new research on the impacts of AI — Google
Google.org announced $15M in new Digital Futures Fund grants for independent research on AI’s societal impact, covering work and the economy, innovation and infrastructure (including AI’s energy demands), and security and governance frameworks.
LARQL — Query neural network weights like a graph database — Lobsters
A Rust-based tool that decompiles transformer models into a queryable “vindex” format, enabling browsing, editing, and recompiling model knowledge without a GPU. Knowledge editing via lightweight JSON patches (~10 KB per fact) avoids full retraining; supports Gemma, Llama, Mistral, Mixtral, and Qwen.
TESSERA — A pixel-wise earth observation foundation model — Lobsters
A foundation model for pixel-level satellite and aerial imagery analysis, enabling granular environmental monitoring, agricultural assessment, and change detection beyond traditional coarse-resolution approaches.
Interviews & Conversations
How does Claude Code actually work? — Theo - t3.gg (39min)
Theo breaks down what an AI coding “harness” actually is — the set of tools and environment in which an agent operates — and builds one from scratch in ~60 lines of Python. Citing Matt Mayer’s independent benchmark showing Opus jumping from 77% in Claude Code to 93% in Cursor, he demonstrates that performance differences come primarily from the harness, not the model. The video shows how tool descriptions in system prompts steer model behavior: marking a tool as “deprecated” in its description causes models to avoid it entirely, and different models (Claude vs. Gemini) respond to the same tool descriptions in meaningfully different ways. Cursor’s edge, he argues, comes from dedicated engineers who micro-adjust system prompts and tool descriptions per model — work Anthropic likely hasn’t invested in for Claude Code.
References
- Stanford report highlights growing disconnect between AI insiders and everyone else — TechCrunch, 2026-04-13 [blog]
- Anti-AI anger hits Sam Altman’s front door — The Rundown, 2026-04-13 [blog]
- An AI Vibe Coding Horror Story — Hacker News, 2026-04-14 [blog]
- What happens when AI runs a retail store — The Rundown, 2026-04-14 [blog]
- Bringing people together at AI for the Economy Forum — Google, 2026-04-14 [blog]
- Google.org and Johnson & Johnson Foundation launch $10M rural healthcare AI initiative — Google, 2026-04-14 [blog]
- UK Department for Transport accelerates policy insights with Google Cloud AI — Google, 2026-04-14 [blog]
- OpenAI has bought AI personal finance startup Hiro — TechCrunch, 2026-04-13 [blog]
- Speeding up GPU kernels by 38% with a multi-agent system — Cursor, 2026-04-13 [blog]
- Supporting new research on the impacts of AI — Google, 2026-04-14 [blog]
- LARQL — Query neural network weights like a graph database — Lobsters, 2026-04-14 [blog]
- TESSERA — A pixel-wise earth observation foundation model — Lobsters, 2026-04-14 [blog]
- How does Claude Code actually work? — Theo - t3.gg, 2026-04-13 [video]