Edition: August 24, 2026

General Intuition, a New York AI startup that builds foundation models for generalized agents in physical space, is seeking a new funding round at a $6 billion pre‑money valuation, backed by investors such as Valor Equity Partners, Point72 Ventures and existing backers Khosla Ventures. The capital will accelerate its robotics focus, expanding compute and talent to push its action‑label‑driven models toward broader task generalization.

OpenAI is expanding its ChatGPT Work platform to let AI agents manage users' digital workflows—from email and Slack to design tools—by granting them deep access to personal data. While this could boost productivity for white‑collar jobs, it raises privacy concerns and highlights the challenge of convincing non‑technical users to adopt agentic AI beyond the engineering niche.

Hugging Face is reportedly in talks to be acquired for $13 billion or more, following interest from investors and banks amid a surge in AI infrastructure deals. The company, valued at $4.5 billion in its last round, says it is close to profitability and is weighing offers while emphasizing its responsibility to the open‑source community.

Children master language with a fraction of the data that massive AI models require, highlighting a stark data‑efficiency gap between human learners and large language models. Researchers argue that understanding how kids learn could lead to far more efficient AI and shed light on fundamental questions about language acquisition.

Instinct, a new AI personal assistant still in private testing, dazzles users with its ability to manage emails, schedules, and other tasks by accessing deep device data, but its Terms of Service grant it perpetual, irrevocable rights to capture, store, and even act on user information, raising serious privacy and security concerns. Early adopters have reported that the bot retained and summarized their emails even after access was revoked, prompting criticism of its data‑handling practices.

A new Stanford study shows AI is disproportionately cutting entry‑level jobs, with workers aged 22‑25 in AI‑heavy roles now 19% less employed than peers in less‑affected fields, up from a 13% gap last year. The findings suggest younger workers are bearing the brunt of automation while older employees remain relatively insulated.

XPENG’s physical AI unit has raised over $900 million at a $6.3 billion valuation, marking the largest single‑round private funding in China’s physical AI sector, to accelerate its IRON humanoid robot platform. Backed by IDG Capital, Tencent and Alibaba, the money will fund full‑stack hardware and software R&D, mass‑production facilities and overseas commercial expansion, even as the parent EV business faces a steep share decline.

Researchers are probing why children learn language far more efficiently than massive AI models, hoping to shrink the data‑efficiency gap and build smarter, leaner systems. Meanwhile, a new niche industry is emerging as entrepreneurs like Roman Chiporukha launch "space travel agent" services, selling multi‑million‑dollar trips and experiences on private missions to the ISS.

A substitute teacher in Los Banos, California, discovered AI‑generated nude images of himself circulating on TikTok, allegedly created by students. The incident sparked a months‑long dispute involving school officials, parents and police, highlighting the distress educators face as deepfake technology enables non‑consensual sexualized content. The case underscores growing concerns over digital harassment and the need for stronger safeguards in schools.

Import AI’s latest issue reviews a METR study showing AI’s uneven impact—dramatic speed‑ups in cybersecurity vulnerability discovery, modest gains in mathematics, and limited effect on core AI research—while also spotlighting new tools like SPADE for auto‑generating synthetic environments and Hawkeye for optimizing GPU kernels. The piece underscores the patchy nature of AI‑driven acceleration across scientific fields and introduces promising infrastructure advances.

Canonical is co‑funding a three‑year PhD project with the University of Bristol to explore using large language models to translate extensive C codebases into safe, maintainable Rust. The research aims to assess whether AI can break down hundreds of thousands of lines of C, rewrite them in Rust, and preserve undocumented legacy behavior, potentially benefiting components like snap‑confine and AppArmor.

The article examines how schools are navigating AI in education, highlighting the mixed reactions of teachers and a case study at Cheshire Academy where staff receive general AI training and use tools like ChatGPT and MagicSchool for lesson planning, while still grappling with concerns over accuracy, bias, and privacy. It also showcases a French teacher’s approach of having students critically evaluate AI-generated edits to understand its limits.

President Donald Trump purchased up to $50,000 of SpaceX stock two weeks after its record‑setting IPO, buying at around $150 per share, which may now be below the IPO price. The purchase highlights Trump’s personal ties to Elon Musk and underscores SpaceX’s growing reliance on government contracts and index inclusion.

The opinion piece praises the 2026 Software Should Work conference as a rare, binge‑worthy series of talks that holistically tackles software reliability—from formal methods and tooling to culture and motivation—offering deep insights across languages, testing, and industry practices. It argues that such focused, high‑quality content is essential for advancing a discipline that underpins the entire digital world.

Replit CEO Amjad Masad will headline the Disrupt Stage at TechCrunch Disrupt 2026 in San Francisco, discussing the future of programming and AI‑driven software creation. The talk comes as Replit’s valuation jumps to $9 billion and its annual run‑rate nears $1 billion, highlighting its role in democratizing code development.