Claude formal verification trial across 123 open-source libraries
The authors report using Claude to produce formal proofs in RISC-V machine code for 239 test entry points across 123 open-source libraries covered by Google OSS-Fuzz. Under specific assumptions, the proofs establish crash-freedom for inputs smaller than 64 MB; they do not prove the absence of all bugs.
The authors say an experiment across 123 libraries shows Claude can scale formal verification for open-source software.
02INDUSTRY
OpenAI discusses raising $30 billion at a $1.25 trillion valuation
According to CNBC, OpenAI is in talks to raise up to $30 billion at a pre-money valuation of $1.25 trillion. Terms are not finalized; the UAE investment may be linked to the company securing more transaction capacity in the region.
OpenAI is discussing raising up to $30 billion in a round led by Coatue and investors in the UAE.
03POLICY & LAW
Meta delays business terms after publisher backlash
After publishers objected, Meta delayed its new business terms from October 30 to November 20. A spokesperson said the delay would allow time to explain the changes and use of web crawlers. Meta says content owners can state preferences via robots.txt; legal experts say removing “training” does not rule out crawling content for AI training.
Meta delayed the start of business terms that raised concerns it might allow news sites to be crawled for AI training until November 20.
04MODELS
Canva unveils Craft, a model that plans designs before generating images
Canva Craft is a two-stage model that first plans design elements and then generates an image. The company says this approach aims to improve accuracy in complex images requiring readable text and precise object placement. The model supports 2K resolution and over 30 aspect ratios; it ranks fourth in Design Arena’s crowdsourced leaderboard.
Canva has launched Craft, an AI image model that plans design elements before generating an image.
05OPEN SOURCE
Berget AI publishes open method for calculating carbon per request
Berget AI published an open-source method for calculating carbon emissions per request. The company estimates that the same Kimi K3 request on the same hardware emits 0.4 grams of CO₂ in Sweden and 3.2 grams in Texas. The method uses grid averages and excludes the water and energy impacts of chip manufacturing.
Berget AI calculates that the same Kimi K3 request emits 0.4 grams of CO₂ in Sweden and 3.2 grams in Texas.
06MODELS
ThinkingCap cuts Gemma-4 reasoning tokens by an average of 38.9%
Bottlecap AI reported that its ThinkingCap version, derived from Google’s Gemma-4-26B-A4B-it, reduced reasoning tokens by an average of 38.9% across 12 benchmarks. The company measured an average accuracy drop of 0.15 percentage points; evaluations used its own test setup.
Bottlecap AI reported that ThinkingCap, based on Gemma-4-26B-A4B-it, cut reasoning tokens by an average of 38.9% across 12 benchmarks, with a measured average accuracy drop of 0.15 percentage points.
✓
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.
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.
