Helping small businesses put AI to work
OpenAI is working with America's SBDC network to offer small businesses more hands-on AI training and local support, and published a report on how small teams use AI.
AI news — 2026-10-04
Daily AI news from official sources, Hacker News and arXiv
Editor's picks
Introducing GPT-6.1 Sol
A cheaper tier close to OpenAI's top model matters for anyone running coding or agent workloads at scale; OpenAI prices it at one-fifth of Astra's standard API token rates.
OpenAI
Towards safety cases for frontier AI training
One of the few public write-ups from a frontier lab on how it intends to argue that a training run is safe, including how misalignment incidents get investigated.
OpenAI
The Agent Said It Was Done. The Database Disagreed.
A useful reminder for agent builders: Microsoft's benchmark checks the backend state an agent leaves behind across repeated runs, not just whether its tool calls look right.
Hugging Face
Editor's note: Summaries were drafted with AI assistance from each publisher's own text and are awaiting human review.
From AI companies & labs
Posts from official company and lab blogs published in the last 7 days.
OpenAI
DevDay 2026 Recap
OpenAI's roundup of DevDay 2026 collects more than 20 announcements spanning GPT-6 Astra, ChatGPT, Codex, its APIs, security and new developer tools.
Introducing GPT-6.1 Sol
OpenAI introduced GPT-6.1 Sol, a model it says comes close to Astra on coding, computer use and professional tasks while costing one-fifth of Astra's standard API token prices.
Introducing dots
OpenAI announced dots, proactive assistants designed to keep working on long-running projects and everyday tasks while users stay in control.
How we will do better for Australia
OpenAI apologised for incidents involving Australian government websites and described stronger safeguards and support for Australia's cyber defences.
Towards safety cases for frontier AI training
OpenAI shared early guidelines for building safety cases around frontier AI training, covering technical safeguards, operational practices and how misalignment incidents are investigated.
Hugging Face
The Agent Said It Was Done. The Database Disagreed.
Microsoft's ThinkingBox, now available through Hugging Face, grades AI agents on the backend state and side effects they leave across 507 business workflows, each run 20 times to measure consistency.
Open-sourcing AstaBrief, the fast report-generation model in Asta
Ai2's Hugging Face post on open-sourcing AstaBrief 8B and its training data explains how the model was trained to turn a research question and retrieved excerpts into a cited report in one pass.
Top AI stories on Hacker News
Stories with AI-related titles, ranked by points over the past 36 hours.
Aleph Alpha Kolibri: How the sovereign German LLM works
OpenAI safety leader quits, warning AI company's culture is 'broken'
Getting the most out of Opus 5.5 in Claude and Claude Code
I quit OpenAI because its culture is broken
LeCun has "zero concerns" about AI wiping out humanity, recent "rogue" incidents
Pop!_OS bans AI-generated code from much of its codebase
US killer's sentence quashed because of AI video of victim shown in court
Religious scholars met with Anthropic
Our AI Midwife
Gemini ending free use of Flash and Pro models
New arXiv papers
Automatic selection of new submissions, preferring papers cross-listed in cs.AI, cs.CL and cs.LG. Not a quality ranking.
No new items.
Sources checked for this edition
- OpenAI: OK (6)
- Hugging Face: OK (2)
- Mistral AI: no new items
- NVIDIA Blog: no new items
- Ai2 (Allen Institute for AI): no new items
- Together AI: no new items
- Qwen (Alibaba): no new items
- Engineering at Meta: no new items
- Berkeley AI Research (BAIR): no new items
- MIT News – Artificial intelligence: no new items
- Hacker News (via Algolia HN Search API): OK (10)
- arXiv cs.AI: no new items
- arXiv cs.CL: no new items
- arXiv cs.LG: no new items
- LiteLLM model_prices_and_context_window.json: OK (747)
Not included:
- Anthropic: No public RSS/Atom feed found (rss.xml, news/rss.xml, feed → 404); not scraped.
- Google DeepMind: Google property — excluded by project policy (only google.com/search allowed).
- Meta AI blog (ai.meta.com): No public RSS feed (404); Engineering at Meta feed used instead (AI-filtered).
- Microsoft Research: Feed returned HTTP 403 via proxy.
- Apple ML Research / AWS ML Blog: apple.com and Amazon properties excluded by project policy.
- Hacker News Firebase API: Hosted on Firebase (Google); Algolia HN Search API used instead.
Updated: 2026-10-04 08:35 UTC