September 23, 2026 |
Artificial intelligence has reached the part of the movie where the scientists address the United Nations while everyone else keeps building the machine. Today’s roundup brings together warnings about powerful AI, international safety talks, another major bet on AI-designed medicine, and the growing business of keeping models running.
Then there is sports, where the question is less “Will AI change civilization?” and more “Can it help us recognize that formation before third down?” Below are six reports published today and three sports features or previews worth your attention. The dates and source types are labeled because a fresh search result does not automatically mean a fresh story.
For more headlines between these roundups, check out The Drone Report.
1. AI leaders take their warnings to the United Nations
Today | September 23
Sam Altman, Dario Amodei and Hugging Face co-founder Clément Delangue briefed the UN Security Council on increasingly capable AI. Reuters reports that the discussion included risks to international security and concerns about systems eventually escaping human control.
These are warnings about potential developments, not confirmation that a rogue superintelligence has arrived. Still, the companies building the technology are asking governments to take its risks seriously. There is a certain Jurassic Park energy to that meeting agenda.
Read: Reuters: AI leaders warn UN of security risks
2. The U.S. and China agree to discuss AI safeguards
Today | September 23
NPR’s John Ruwitch reports through WBUR that the United States and China have agreed to talks about safeguards for AI development. The report emphasizes the obstacles: longstanding mistrust and different priorities.
Agreeing to talk is a starting point. It does not establish an enforceable agreement or prove that either country will slow development. My read: even a limited conversation matters when the competitors building powerful systems also need ways to communicate when something goes wrong.
Think less “world peace achieved” and more “we should probably exchange emergency numbers.”
Read/listen: WBUR: China and the U.S. agree to talk about AI safeguards
3. AI drug development gets another $140 million vote of confidence
Today | September 23
Basecamp Research raised $140 million in a Series C led by S32. Participants included Nvidia’s NVentures and the Anthology Fund, a partnership between Anthropic and Menlo Ventures, according to Tech.eu.
The company uses AI and biodiversity data to design potential treatments. It says the money will support new models and advance its therapeutic pipeline, including work involving the reprogramming of cells inside the body.
The distinction matters: funding promising research is not the same as proving a medicine works. Biology still gets a vote, regardless of how impressive the pitch deck looks.
Read: Tech.eu: Anthropic and Nvidia back Basecamp Research
4. Thinking Machines signs a $65 million annual computing deal
Today | September 23 | Company announcement
Crusoe announced a $65 million annual agreement to run open-model inference workloads for Thinking Machines Lab. Inference is the computing involved when a trained model processes a request and produces an output.
That makes this a useful reminder about the economics behind the chatbot window. Building a model is only part of the expense. Serving it reliably to users is another continuing bill.
Crusoe says its managed service will provide dedicated infrastructure and operational support. The companies also describe potential future work on generating synthetic data. Those plans should be read as plans, rather than completed results.
The cloud, it turns out, accepts neither exposure nor positive vibes as payment.
Read: Crusoe’s partnership announcement
5. OpenAI is expanding the people who teach AI skills
Today | September 23 | Company announcement
OpenAI marked two years of its Academy by announcing a pilot Community Trainer Program. Participating organizations nominate staff to learn its curriculum, complete training and a facilitation assessment, and then lead practical workshops.
OpenAI also says it has hosted more than 250 Academy events and that more than four million people have engaged with its content. Those are company-reported participation figures, not independent measurements of learning outcomes.
The useful angle is local instruction for schools, workforce groups, nonprofits and small businesses. Access to a chatbot is easy to confuse with knowing how to use one well. They are very different skills, especially when the answer arrives confidently and incorrectly.
Read: OpenAI: Two years of OpenAI Academy
6. Tax professionals want AI accountability before the paperwork gets weird
Today | September 23
The AICPA and former IRS commissioner Danny Werfel are leading the Council on AI Risk in Tax. The Tax Adviser reports that the initiative brings together participants from accounting, law, technology, government and academia.
Its work includes refining a framework covering information integrity, fairness, security and organizational capacity. The council also aims to develop practical tools and guidance.
This is one of the less cinematic AI stories, but it deals with a very real problem: who checks the work when software becomes part of a consequential decision? “The chatbot sounded pretty sure” is a weak quality-control system.
Somewhere, a spreadsheet has just requested adult supervision.
Read: The Tax Adviser: New initiative on AI in tax
7. Sports spotlight: What the Seahawks are actually doing with AI
Earlier coverage | September 9 | Microsoft feature
Microsoft’s feature on the NFL’s Game Analytics Dashboard describes how Seattle and other teams use Copilot in Excel to help analyze football data. For the 2026 season, updates include AI-assisted data editing and the ability to send live tables and charts to Sideline Viewing System devices.
One example involves analyzing run/pass tendencies against particular personnel groupings. The point is to help analysts get useful information to coaches faster.
This is a more concrete story than the idea of a chatbot secretly running an offense. The feature describes analysts using tools to support coaching decisions; it does not establish that an AI independently calls plays or wins games.
As a Seahawks fan, I appreciate any spreadsheet willing to put in the film work. It still has to survive third-and-long like everyone else.
Read: Microsoft: How NFL teams use the Game Analytics Dashboard
8. Sports spotlight: AI is helping build the schedule everyone will complain about
Featured video | Publication date not displayed on the page
A Front Office Sports episode features Fastbreak AI CEO John Stewart and chief product officer Chris Groer discussing scheduling technology. The conversation covers competing demands such as travel, broadcasting, league operations and competitive balance.
It is an overlooked application with an understandable purpose. A sports schedule must satisfy far more conditions than “have these two teams play on Sunday.”
The feature is a discussion with the technology’s vendors, not an independent audit of better outcomes. But it offers a useful look at why leagues pursue these systems.
Fans will still conclude that the schedule personally hates their team. Some traditions are beyond the reach of automation.
Watch: Front Office Sports: The hidden AI technology building pro sports schedules
9. Sports watchlist: Next week’s AI conversations go well beyond the scoreboard
Upcoming event | September 29–30
SportsPro AI + Tech is scheduled for London Stadium next week. Its published program includes representatives from organizations such as Nvidia, Formula One, the NHL, UEFA and WTA Ventures.
Topics include AI governance, fan data, broadcast production, personalized experiences and agent-based workflows. These are scheduled discussions, not new product announcements.
The agenda is worth watching because it shows how widely sports organizations are considering AI. Coaching is only one piece. Content, ticketing, operations and marketing are part of the conversation too.
My question for all of it: does this make being a fan better, or just make selling things to fans more efficient? Those outcomes can overlap. They are not interchangeable.
Explore: SportsPro AI + Tech agenda and speakers
Keep an eye on The Drone Report for more AI headlines throughout the day.
Suggested WordPress slug: ai-news-daily-september-23-2026-sports-un
SEO description: AI leaders address the UN, new drug research draws funding, and sports organizations put AI to work in analytics, scheduling and fan experiences.
Suggested category: AI News Briefs
Suggested tags: AI news, artificial intelligence, NFL, Seattle Seahawks, sports technology, Anthropic, Nvidia, OpenAI
Leave a Reply