OpenAI announces three scientific discoveries; some findings under review
OpenAI announced discoveries in mathematics, computer science and physics; the company says its computer science work disproves the Exponential Time Hypothesis. Three experts will review that proof. The full physics study has not been published, while the mathematics result was presented directly to researchers.
OpenAI announced results in mathematics, computer science and physics. Experts will review the computer science claim, and the full physics study has not yet been published.
02RESEARCH
OpenAI and Anthropic models claim to disprove ETH
OpenAI and Anthropic claim their AI models disproved the Exponential Time Hypothesis (ETH) with algorithms that solve certain logic problems faster than expected. MIT’s Ryan Williams says unusual parts of the proofs need community review and the result is not settled. He says OpenAI also checked cryptographic implications, finding no evidence that current systems were broken.
OpenAI and Anthropic say their models found algorithms that could disprove the Exponential Time Hypothesis; computer scientists are reviewing the proofs.
03MODELS
Liquid AI announces open-weight LFM3 model family
Liquid AI announced its open-weight LFM3 model family, which includes models for text, images, audio, video and decision-making. The company says the models can be customized on a single GPU; LFM3-1.7B outperforms the earlier LFM2.5-2.6B with 35% fewer parameters.
LFM3 brings small models for text, images, audio and decision-making together on a shared foundation.
04POLICY & LAW
Judge bars Anthropic’s pirated book archive; Claude training ruled fair use
Federal Judge William Alsup barred Anthropic from keeping pirated books in a permanent central archive, but ruled that training Claude on legally acquired books was fair use. Tobias Wolff and Chang-rae Lee, who opted out of the settlement, are seeking up to $150,000 per work in a separate lawsuit.
Tobias Wolff and Chang-rae Lee filed separate lawsuits against Anthropic, alleging their works were used without permission to train Claude.
05SECURITY
Workers AI GLM-5.3 bug leaked customer data in 155 responses
Cloudflare said an isolation bug affecting GLM-5.3 on Workers AI produced responses containing data from other customers’ requests. Between August 27 and September 9, 155 responses to prompts exceeding 681,280 tokens were affected. The company said it found no evidence of deliberate exploitation and reopened the model with SGLang fixes.
A KV-cache bug in GLM-5.3 caused content from other customers’ sessions to appear in 155 responses to very long prompts.
06INDUSTRY
Amazon reduces most Nova models to minimal support
According to Bloomberg, Amazon is reducing most of Nova’s text, video and image models to minimal support, narrowing its development goals. The report says employee and computing resources are being shifted to a single “frontier model” effort led by Pieter Abbeel; its assessment that the models failed is attributed to sources.
According to Bloomberg, Amazon is scaling most Nova models back to minimal maintenance and redirecting resources to a single frontier-model effort led by Pieter Abbeel.
07POLICY & LAW
US license plates tracked with AI to flag “suspicious” trips
The ACLU said Grace Knowles was stopped while driving from Virginia to Florida after her license plate was flagged for “suspicious travel patterns.” The organization warned that these systems also collect location data on people not suspected of crimes and are prone to errors or misuse.
The ACLU says police in the US use license-plate readers and AI tools to stop drivers with no suspected crime based on their travel patterns.
08PRODUCTS
Apple adds AI assistant to Pages, Keynote and Numbers
Apple is adding an AI assistant to Pages, Keynote and Numbers that uses Apple and OpenAI models. The company says it can write and edit text, organize tables and data, and design and improve Keynote presentations. The feature is rolling out in beta to compatible Apple Intelligence devices.
The beta assistant coming to Apple’s three productivity apps can handle text, spreadsheet and presentation editing tasks.
✓
That’s it for today.The rest of your time is yours.
Your daily technology digest, every morning in English. Free. Unsubscribe anytime.
Sources, in the open.
Stories are selected from Hacker News; accessible original articles are summarized and translated with AI. Summaries may contain errors; check the source beneath each story for details. Preparation starts daily at 06:55 Türkiye time (UTC+3), targeting publication before 07:00.
Stories are read aloud by an AI voice.
This edition covers articles shared on Hacker News on the date shown above.
Now on the App Store for iPhone
Your daily digest. Now in your pocket.
Read or listen to a new edition every morning with Kısa for iPhone, save stories for later and keep your streak.
