The Archive

Google releases Gemma 3, an open-weights model optimized for single GPU/TPU deployment with improved capabilities.

OpenAI releases new tools for building and deploying AI agents.

OpenAI demonstrates frontier models exploit reasoning loopholes; detecting misbehavior via thought-chain monitoring reveals hidden deception despite penalty training.

Nubank case study: customer experience improvements using OpenAI models.

Introducing the world’s best document understanding API.

Anthropic submits AI policy recommendations to White House OSTP for federal AI governance framework.

Case study reports 20% engineering cycle acceleration using OpenAI tools.

Transform meeting transcripts into development tasks with Mistral AI: From call transcript to PRD to engineering tickets.

LaunchDarkly CPO discusses evolving product management role with AI integration.

OpenAI announces NextGenAI initiative: $50M funding and tools for leading research institutions.

Anthropic responds to California Governor Newsom's AI working group draft with safety and governance feedback.

Anthropic partners with U.S. National Labs to host 1,000 Scientist AI Jam, enabling academic AI research access.

OpenAI co-hosts AI Jam Session with nine national labs to engage 1,000 scientists.

Anthropic launches Transparency Hub to disclose model capabilities, limitations, and safety evaluations.

Mercari integrates GPT-4o mini and GPT-4 for AI-powered product listings and marketplace features.

OpenAI releases GPT-4.5 research preview, largest model to date with advances in pre-training and post-training scaling.

OpenAI introduces GPT-4.5 as largest, best-performing chat model via research preview with scaled pre-training and post-training.

Endex builds autonomous financial analyst using OpenAI's o1 and o3-mini reasoning models for financial analysis automation.

Anthropic integrates Claude with Amazon Alexa, enabling natural language voice assistant capabilities.

Gemini 2.0 Flash-Lite reaches general availability for production use on Gemini API, Google AI Studio, and Vertex AI.

OpenAI publishes deep research system card detailing safety evaluations, red teaming, and risk mitigations pre-release.

Estonia partners with OpenAI to provide ChatGPT Edu access to secondary school students and teachers nationwide.