openai proposes 5% equity gift to us wealth fund
openai ceo sam altman suggested donating 5% of the company's equity to a proposed us sovereign wealth fund, with other ai firms possibly contributing similar stakes.
author
baris
Baris operates SummarizedData and maintains its sources, publishing rules, and automation. Read the editorial process.
humanity's last exam is a distraction
a look at the hle benchmark, why it was created, and the divided expert opinions on its value for evaluating ai systems.
filtered mixture-of-generators for synthetic survival data
fogs selects plausible synthetic samples from multiple tabular generators to improve downstream survival model training on small clinical datasets.
fourier neural operators learn pde solutions efficiently
new theory proves fourier neural operators can approximate and learn time-dependent solutions of dissipative partial differential equations with polynomial sample complexity.
deep multitask learning for mixed outcomes with shared sparsity
a new method uses deep neural networks and rank-based optimization to handle mixed outcome types in multitask learning with shared predictor selection.
bounded morality: a computational framework for finite agents
a new framework models moral cognition as a tradeoff between moral breadth and depth under limited resources, recasting ethical theories as efficient strategies.
semiscope: tuning gains in semi-supervised security classification
a study disentangles whether improvements in semi-supervised learning for security come from tuning the classifier alone or from joint optimization with the ssl pipeline.
ai alignment must handle shifting human preferences
a new framework treats alignment as controlling how preferences evolve through interaction, not just satisfying fixed goals.
indian founder bets $30m on ai-first office suite
bhavin turakhia self-funds neo, an enterprise platform rebuilt from scratch for ai, challenging microsoft and google.
grpo, dr. grpo, and dapo are three settings of one dial
three popular language model training methods all adjust the same number: the standard deviation of correctness marks across sampled answers.
hugging face and cerebras speed up voice ai with gemma 4
hugging face and cerebras demonstrate a real-time speech-to-speech pipeline using gemma 4, cutting latency for more natural voice interactions.
daily brief: 2026-07-01
google rolls out gemini 3.5 live translate and a mac agent, cloudflare blocks mixed-use ai crawlers, and meta plans a cloud business for excess compute.
google ai updates from june 2026
google announced gemini 3.5 live translate, android 17 features, and a new home speaker, plus tools for research and education.
cloudflare blocks mixed-use ai crawlers from ad-supported sites
cloudflare will default to blocking crawlers that combine search, ai training, and agent use from pages with ads, pushing ai companies to pay publishers.
shopify joins pytorch foundation as platinum member
shopify becomes a platinum member of the pytorch foundation to invest in open source ai for commerce.
meta plans cloud business to sell excess ai compute
meta is developing a cloud infrastructure business to sell access to ai compute and models, following spacex's lead.
when feedback actually helps ai improve
a controlled study finds that multi-turn improvement often comes from resampling or extra computation, not from feedback itself, with only strong external teachers providing real gains.
gemini spark agentic assistant arrives on mac
google's gemini spark desktop agent launches on mac with new app integrations and real-time tracking.