AI Daily — September 12, 2026
Models & Research
GPT-6 Astra takes on greater autonomy at Perplexity and Cognition — Perplexity reports using GPT-6 Astra to draft communications, modify software, and monitor production systems, with less frequent human check-ins than with earlier models. Separately, Cognition is using Astra to improve Devin’s ability to test its own work, aiming to reduce manual code review. OpenAI ↗
My takeaway: The operational shift is less frequent human oversight. You need to account for how much damage a mistake could cause before someone catches it.
Policy & Society
Mathematicians escalate criticism of AI companies amid OpenAI dispute — Twenty-five prominent mathematicians have signed an open letter warning that AI companies’ race to solve mathematical problems threatens attribution and the development of mathematical understanding. The letter comes amid an ongoing dispute with OpenAI. TechCrunch ↗
My takeaway: The real concern is that inputs to coding assistants may feed vendor training and competitive advantage.
Anthropic researcher resigns over AI safety fears, sparking debate — Jacob Coxon says he resigned from Anthropic because AI companies are racing toward self-improving systems without adequately addressing the risks. His warnings have renewed debate over whether advanced AI could threaten humanity, with experts divided between calls for urgent safeguards and skepticism about dramatic extinction predictions. TodayInCanada ↗
My takeaway: I agree that the speed of AI evolution is way too fast, and we need to focus more on setting up proper guardrails to mitigate irregular AI behaviour
Industry & Funding
- Robot-data startup Mecka AI closing in on $500M valuation
- Moonshot AI eyes $2 billion in yearly revenue with Kimi models
- Nscale recruits ex-OpenAI executive Fidji Simo as it eyes IPO
NVIDIA combines Palantir Foundry and cuOpt to manage chip supply allocation — NVIDIA is using Palantir’s Foundry platform and its own cuOpt solver to automate hardware allocation across global manufacturing sites, tracking delivery from wafer output to the first token generated in a data centre. AINEWS ↗
My takeaway: The useful pattern is the division of work. A solver handles supply constraints, while a smaller fine-tuned model interprets unstructured information. An ontology connects supplier, inventory, and production targets within a shared model of the supply chain.
Summaries are AI-generated and may contain errors — always verify against the linked original. Each story links to its source, which holds the copyright. Outlet names are shown for attribution only and do not imply any endorsement or affiliation.
Disclaimer: The views expressed in My Takeaway are my own personal opinions and general observations on industry trends. They are not intended to criticize, disparage, or make factual claims about any specific company, product, or platform. Any platform names mentioned are referenced solely for illustrative and informational purposes.