OpenAI, Anthropic Warn UN: AI Crossroads

Key Takeaways

• OpenAI and Anthropic CEOs warned at the UN that humans could lose control of AI systems and that the world is at a crossroads regarding AI development.
• Mistral AI director Arthur Mensch believes AI is controllable software, contrasting with broader warnings about AI risks following the hacking of OpenAI software.
• Congress is advised to lead a public fact-finding mission before creating AI policy, focusing on what specific technologies concern leaders like Sam Altman and Dario Amodei.
• Google, OpenAI, and Anthropic plan to launch a new AI safety standards body by late 2026 or early 2027 without government oversight.
• Museums are using AI to improve donor research and fundraising, with the Frye Museum seeing a 20% increase in repeat web traffic after adopting AI for consistent messaging.
• Meta unveiled 100-gram VR glasses featuring a 5K micro-OLED display with 37 pixels per degree and ultra-compact pancake lenses.
• WhalesBot launched EliteMind, a modular Physical AI robot powered by W-Hong LLM that connects to the OceanLevel learning platform for hands-on STEM education.
• AI tools are reshaping outpatient care facility design by testing clinic layouts and simulating patient flow, though experts stress architects and clinicians must remain in control.
• Economist Jason Furman says the U.S.-China AI race may have no clear winner and calls for dialogue and cooperation on AI safety.
• AI automation risks increasing inequality rather than creating abundance, with benefits flowing to those who own machines and data, mirroring patterns from the Industrial Revolution.

Museums share AI strategies for nonprofit fundraising

A report released in April shows how art museums use AI to improve donor research and fundraising. AI helps museums find potential donors and personalize outreach through emails and letters. Traditional methods like phone calls are slow, but AI works faster and more accurately. The report offers lessons for nonprofits wanting to use AI. It is available on the American Alliance of Museums website.

Report highlights five ways AI can help museums connect with people

A report by Remuseum titled Five Ways that AI Can Humanise Museums explores AI use in US art institutions. It draws on a January 2026 meeting with representatives from seven museums including MoMA. The five uses include art and technology residencies, fundraising, communication, staff training, and personalized visitor experiences. MoMA launched an AI-powered digital wall in 2022. The Frye Museum saw a 20% increase in repeat web traffic after using AI for consistent messaging. The report stresses that AI needs ethical guidelines and cannot replace the experience of viewing art in person.

OpenAI and Anthropic CEOs urge global AI cooperation at UN

The CEOs of OpenAI and Anthropic spoke at the UN about the need for global cooperation on AI. They warned that humans could lose control of AI systems. They also warned that AI systems could fall into the wrong hands. The leaders said the world is at a crossroads regarding AI development.

Mistral AI CEO says artificial intelligence can be controlled

Arthur Mensch is the co-founder and director general of French start-up Mistral AI. He believes AI is software and can be controlled. His comments come amid warnings about AI risks including the prospect of an AI-pocalypse. These warnings have grown since OpenAI software was hacked. The interview took place in Mistral AI's Paris offices on September 22, 2026.

Opinion piece advises Congress on questions to ask AI leaders

An opinion piece addresses Congress on what to ask AI leaders like Sam Altman and Dario Amodei. The author says Congress does not yet understand enough about AI labs to create effective responses. The writer advises Congress to lead a public fact-finding mission first. The author suggests focusing questions on three key areas. One key area is understanding what specific types of AI technology concern leaders. The piece explains that AI is a broad category including self-driving cars and chatbots. The author notes that leaders may actually be referring to large language model-powered agents when discussing AI dangers.

Meta reveals 5K micro-OLED VR glasses with pancake lenses

Meta showed a pair of VR glasses weighing 100 grams with a 5K micro-OLED display. The display has 37 pixels per degree for clear text. The glasses use ultra-compact pancake lenses and are designed for tasks like streaming movies and working on an infinite screen. Full specs on weight, battery life, and performance are not yet available.

Google OpenAI Anthropic plan AI safety standards group

Google, OpenAI, and Anthropic are working together to create a new AI safety standards body. The group plans to launch by late 2026 or early 2027 without government oversight. The companies want to develop shared guidelines for safe and responsible AI development. They also hope to include other organizations in a collaborative process.

WhalesBot launches EliteMind robot for AI education

WhalesBot launched EliteMind, a modular Physical AI robot for STEM and AI education. Powered by W-Hong LLM, it lets students move AI models from digital learning into real-world sensing and action. EliteMind connects to OceanLevel, WhalesBot's learning platform, for a full cycle of lessons and hands-on practice. The robot supports voice interaction, visual understanding, and physical control.

AI is reshaping outpatient care facility design

AI tools are changing how outpatient care buildings are planned and designed. Generative tools can test thousands of clinic layouts and simulate patient flow before construction. Digital twins and responsive rooms are expected to improve efficiency and caregiver time. Experts stress that architects and clinicians must stay in control to protect privacy, equity, and safety.

U.S. China AI race may have no clear winner

The U.S. and China are investing heavily in AI, but economist Jason Furman says a clear winner is not guaranteed. He says being first matters in some areas like cybersecurity, but broader AI benefits may depend less on who leads. Furman also calls for dialogue and cooperation on AI safety. He warns that knowledge in this field will spread regardless of competition.

AI automation may increase inequality not abundance

New technology and AI automation will not create an age of abundance for everyone. The benefits of new tools will go to those who own the machines and data. Workers who lose their jobs to automation may be left behind. This pattern mirrors the Industrial Revolution. Solving these problems may require new economic systems and ideas like universal basic income, but funding and effectiveness remain unclear.

Sources

• Museums provide lessons on how to use AI effectively
• AI can strengthen human connections to museums, report suggests
• OpenAI, Anthropic CEOs at UN call for global cooperation on AI: 'We are at a crossroads'
• Arthur Mensch, CEO of French start-up Mistral AI: 'AI is software. It can be controlled'
• Opinion | Dear Congress: Here’s What to Ask A.I. Leaders
• Meta shows VR glasses with 5K micro-OLED display
• Google, OpenAI and Anthropic AI Safety Group Takes Shape
• WhalesBot Launches EliteMind, Bringing Physical AI Into STEM & AI Education
• Designing for Intelligence: AI’s Emerging Role in Outpatient Care Environments
• Who’ll win AI race? U.S.? China? Maybe no one.
• Beyond the AI abundance delusion