The Archive
Search the full wire by company, model, lab, or keyword. Every story we have ever aggregated.
Sycophancy in GPT-4o: what happened and what we’re doing about it
OpenAI rolls back GPT-4o ChatGPT update due to excessive sycophancy; restores earlier balanced behavior.
Introducing the Anthropic Economic Advisory Council
Anthropic establishes Economic Advisory Council to guide AI's societal and economic impact.
Anthropic Economic Index: AI’s impact on software development
Claude Code achieves 79% automation in coding vs. 49% on Claude.ai; startups adopt agentic tools 3x faster than enterprises.
Music AI Sandbox, now with new features and broader access
Google Music AI Sandbox expands feature set and access for music professionals exploring generative AI.
New in ChatGPT for Business: April 2025
OpenAI announces ChatGPT for Business updates: o3 model, image generation, memory, internal knowledge integration.
Detecting and countering malicious uses of Claude: March 2025
Anthropic releases March 2025 report on detecting and mitigating malicious Claude misuse.
Introducing our latest image generation model in the API
OpenAI releases gpt-image-1 image generation model via API for developer integration.
Speak is personalizing language learning with AI
Speak CEO discusses AI-powered language learning personalization in conversation format.
The Washington Post partners with OpenAI on search content
Washington Post integrates news content into ChatGPT via partnership for summaries and citations.
Our approach to understanding and addressing AI harms
Anthropic outlines framework for identifying, measuring, and addressing harms from AI systems.
Introducing Gemini 2.5 Flash
Google launches Gemini 2.5 Flash, first fully hybrid reasoning model enabling developers to toggle thinking on or off.
Introducing OpenAI o3 and o4-mini
OpenAI releases o3 and o4-mini models with reasoning, tool access, web browsing, code execution.
Thinking with images
Brief announcement on multimodal reasoning capabilities combining images with analysis.
OpenAI o3 and o4-mini System Card
OpenAI o3 and o4-mini system card details tool capabilities, reasoning, and safety features.
Generate videos in Gemini and Whisk with Veo 2
Google releases Veo 2 video generation in Gemini and Whisk, creating eight-second high-resolution videos from text or image prompts.
OpenAI announces nonprofit commission advisors
OpenAI appoints four nonprofit advisors to guide philanthropic strategy.
Our updated Preparedness Framework
OpenAI updates Preparedness Framework for measuring and mitigating frontier AI severe harm risks.
DolphinGemma: How Google AI is helping decode dolphin communication
Google applies Gemini LLM to dolphin communication analysis through DolphinGemma, a domain-specialized model variant.
Introducing GPT-4.1 in the API
OpenAI releases GPT-4.1 API family with coding, long-context, and instruction-following improvements plus nano model.