OpenAI Releases GPT-5.2, Google Powers US Military AI Platform
OpenAI launches GPT-5.2 for agentic AI applications while Google supports new military platform GenAI.mil and Trump administration moves on AI regulation.
Sources · Last Week in AI
The news desk
Recent developments in artificial intelligence, with the original sources one click away.
OpenAI launches GPT-5.2 for agentic AI applications while Google supports new military platform GenAI.mil and Trump administration moves on AI regulation.
Sources · Last Week in AI
OpenAI has rolled back its model router feature for most ChatGPT users following concerns about service quality and past user complaints.
Sources · WIRED AI
Three new research papers examine different approaches to adapting and editing large language models while addressing resource constraints and knowledge preservation.
Sources · arXiv AI (cs.AI)
Zoom announced a record score on a major AI test, drawing skepticism, while Zencoder launched a free tool comparing Claude and OpenAI models for coding.
Sources · VentureBeat AI / arXiv AI (cs.AI)
Motif, a Korean AI startup, reveals lessons on enterprise LLM training, while new research explores deception in multi-agent AI systems.
Sources · VentureBeat AI / arXiv AI (cs.AI)
Three arXiv papers examine LLMs for retail forecasting monitoring, clinical guideline queries, and inequality theorem proving.
Sources · arXiv AI (cs.AI)
NVIDIA releases guide on fine-tuning LLMs with Unsloth, while researchers compare classification approaches for resource-constrained environments.
Sources · NVIDIA AI Blog / arXiv AI (cs.AI)
Hannah Wong, OpenAI's Chief Communications Officer, announced her departure as the company launches an executive search for her replacement.
Sources · WIRED AI
Recent arXiv preprints tackle autonomous vehicle data curation, reinforcement learning for reasoning models, and supervised fine-tuning improvements.
Sources · arXiv AI (cs.AI)
Large language models have demonstrated human-expert level abilities in analyzing language structure and patterns, marking a significant milestone.
Sources · WIRED AI
Three new research papers introduce automated frameworks for MDP modeling, epidemic response planning, and zero-shot action recognition.
Sources · arXiv AI (cs.AI)
Three arXiv papers examine attention mechanisms in LLM reasoning, zero-shot 3D map generation with LLM agents, and improved reinforcement learning methods.
Sources · arXiv AI (cs.AI)
Researchers release frameworks addressing reliability, scalability, and constraint satisfaction in LLM-based multi-agent systems.
Sources · arXiv AI (cs.AI)
VentureBeat explores how AI technology is disrupting the traditional build versus buy decision-making process for software solutions.
Sources · VentureBeat AI
Most enterprise AI coding deployments underperform due to implementation issues rather than model limitations, according to VentureBeat analysis.
Sources · VentureBeat AI
Google is integrating Gemini's advanced translation capabilities into Google Translate, alongside additional new features.
Sources · Google AI Blog
OpenAI reveals its Codex AI coding agent is now used to develop the majority of its own codebase, according to company statements.
Sources · Ars Technica AI
TechCrunch publishes detailed resource tracking ChatGPT's product updates and releases throughout the year.
Sources · TechCrunch AI
OpenAI releases GPT-5.2 amid competition from Google's Gemini 3, while researchers introduce LabelFusion for combining LLMs with traditional classifiers.
Sources · VentureBeat AI / arXiv AI (cs.AI)
OpenAI's GPT-5.2 achieves state-of-the-art results on math and science benchmarks, reportedly solving an open theoretical problem.
Sources · OpenAI