openai OpenAI and Anthropic Agree to US Government AI Safety Testing OpenAI and Anthropic sign first-of-their-kind agreements with the U.S. AI Safety Institute to evaluate their models for potential risks before and after public release.
openai OpenAI Introduces CriticGPT to Catch AI Hallucinations OpenAI unveils CriticGPT, a new model designed to review and find errors in code generated by ChatGPT. This AI critic aims to solve the growing problem of hidden hallucinations as language models become more advanced.
openai OpenAI Forms Internal Safety and Security Committee Ahead of New AI Model The OpenAI Board creates a new Safety and Security Committee to evaluate and improve safeguards across all projects within 90 days. Security experts praise the proactive move but caution that the lack of independent oversight could lead to an echo chamber.
openai OpenAI Safety Leader Resigns Over Shiny Products Taking Priority OpenAI's Jan Leike resigns and claims the company prioritizes flashy products over crucial AI safety measures. His departure follows the sudden exit of co-founder and chief scientist Ilya Sutskever.
openai OpenAI Forms Safety Committee Ahead of Next Frontier AI Model Release OpenAI establishes a new safety and security committee to evaluate safeguards as it begins training its next frontier model. The move follows recent high-profile departures from the company's safety-focused teams.
openai OpenAI Drops Restrictive Nondisparagement Clause Tied to Employee Equity OpenAI officially scraps a controversial exit agreement that forces departing employees to choose between their vested equity and their right to criticize the company. CEO Sam Altman expresses embarrassment over the provision as the AI firm faces multiple public relations challenges.
anthropic Anthropic Maps Millions of Concepts Inside Claude Sonnet Anthropic uses dictionary learning to identify how concepts are represented inside the Claude Sonnet AI model. This breakthrough opens the black box of a production-grade language model to improve safety.
google Google Unveils Veo Video and Imagen 3 Image Generators Google introduces Veo, a high-definition video generation model, and Imagen 3, an advanced text-to-image tool, both developed alongside artists. The company emphasizes responsible deployment through safety filters and digital watermarks.
ai safety UK and US Form Historic Partnership to Test Advanced AI Models The UK and US sign a Memorandum of Understanding to align their scientific approaches and develop shared safety evaluations for advanced artificial intelligence.
sora New Research Paper Analyzes OpenAI's Sora Video Generation Model A comprehensive review explores the underlying technology, applications, and limitations of OpenAI's text-to-video AI model Sora. The study highlights both the potential impact across various industries and the safety challenges that remain.
google Google's Gemini Fumbles Image Generation With Overly Strict Diversity Filters Google faces backlash after its Gemini AI produces historically inaccurate images due to excessively strict internal diversity filters. The company temporarily disables the chatbot's ability to generate images of people while it investigates the issue.
openai OpenAI Unveils Sora, a Powerful Text-to-Video AI Tool OpenAI introduces Sora, a new artificial intelligence tool that generates high-definition videos up to a minute long from simple text or image prompts. The tool remains restricted to select researchers and creators while the company addresses remaining bugs and implements safety measures.
openai OpenAI Board Expansion Ends Investigation, Altman Regains Control OpenAI concludes its internal investigation with no findings of wrongdoing and adds four new members, including Sam Altman, effectively ending the recent leadership crisis.
openai OpenAI Crisis Reveals Clash Between Safety Guardians and Tech Visionaries An analysis of the November 2023 OpenAI board crisis shows that 95% of employees threatening to quit stems from a collision of hidden assumptions rather than a simple failure of rationality.
openai OpenAI Boardroom Drama Starring Andrew Garfield Gets Film Adaptation The chaotic 2023 ouster and swift return of OpenAI CEO Sam Altman is becoming a movie directed by Luca Guadagnino. The film highlights the deep ideological clash over AI safety between Altman and co-founder Ilya Sutskever.