OpenAI AI Updates: September 15, 2026

1. OpenAI Acquired Glass Imaging for More Than $300 Million

OpenAI. The Wall Street Journal reported the acquisition of Glass Imaging, a computational photography company founded in 2019 by Ziv Attar and Tom Bishop, both former Apple engineers who led the team behind Portrait Mode. Glass Imaging uses neural networks to improve image quality at capture time rather than in post-processing, working around the physical constraints of smartphone sensor size. The company had raised roughly $30 million before the deal. The acquisition fits OpenAI’s reported hardware roadmap, which includes phones, earbuds, and companion devices. Source

2. OpenAI Published a Fyxer Case Study on Fine-Tuning for Voice Matching

OpenAI. The case study covers how Fyxer built an AI executive assistant that organizes inboxes and drafts email in each user’s own voice, combining OpenAI models with fine-tuning, memory, and a real user feedback loop. The emphasis is on trust as the product constraint: the assistant has to produce drafts users will send without rewriting. Source

3. OpenAI Now Runs Safety Checks Before Major Training Runs

OpenAI. Sam Altman reiterated support for pacing AI development while saying rapid progress will continue. According to the report, OpenAI has added safety checks ahead of major training runs, and The Information reports the company has been in discussions with Anthropic and Google for months about joint self-regulation. The position follows Dario Amodei’s call to slow the rate of capability gains. Source