September 25, 2026 | AI, data, world affairs, and a little Tennessee orange
Artificial intelligence is having another quiet week. By quiet, I mean regulators are asking who takes the blame when an AI agent causes trouble, a Federal Reserve official is questioning the size of the AI economy, and tech companies are still placing enormous bets on what comes next. Somewhere in the middle of all that, businesses would like to know whether the chatbot is actually saving them money.
Today’s roundup covers the latest developments alongside a few important stories from earlier this week. There is also a college football connection and a Knoxville update as President Donald Trump prepares to attend Tennessee’s game against Texas. For more headlines between these roundups, visit The Drone Report.
1. The FTC is not buying the “AI did it” excuse
Speaking in Austin on September 25, FTC Chairman Andrew Ferguson pushed back on descriptions of AI agents as independent beings with their own desires. Reuters reports that he suggested developers who direct the tools could face responsibility for resulting harm. He also pointed to existing enforcement powers, including those concerning undisclosed data breaches.
My take: This is an accountability argument, not an announcement that every AI liability question has been settled. Businesses handing software permission to act should expect questions about instructions, access, and oversight. “The robot went rogue” is a great movie trailer. It is a less convincing incident report.
Read the Reuters report via MarketScreener.
2. Is the AI economy becoming too big to fail?
Kansas City Federal Reserve President Jeff Schmid said on September 25 that policymakers need to understand whether the expanding network of AI companies, contracts, and data centers could create systemic financial risks. His remarks raised a question about the industry’s interconnectedness rather than declaring that an AI bailout is coming.
My take: The interesting story is no longer just which company builds the smartest model. It is how much of the wider economy becomes dependent on the buildout working as advertised. Apparently, the sequel to The Big Short may come with a GPU.
Read the Reuters report via Kitco.
3. Trump and tech leaders reportedly have another AI meeting ahead
Reuters reported on September 24 that Trump, House Speaker Mike Johnson, and technology CEOs were expected to meet about artificial intelligence on September 29. The report cited a source familiar with the plans and said further details were not immediately available.
My take: The meeting is worth watching, but the agenda and outcome should not be invented in advance. The useful questions are whether officials address oversight, security, and concrete responsibilities. A room full of powerful people is a photo opportunity until it produces something measurable.
Read the Reuters report via Knoxville’s B97.5.
4. China’s AI race includes both bigger models and homegrown chips
Alibaba announced plans this week to train a model with 5 trillion to 10 trillion parameters and unveiled a new AI chip. Reuters placed the announcements in the wider effort by Chinese companies to develop alternatives to Nvidia hardware amid U.S. export restrictions. These are development plans and company claims, not independent proof of superior performance.
My take: Model size makes a dramatic headline. Useful results, reliability, and the cost of running the system are the tests that matter to customers.
5. The data behind AI is attracting serious money
Snorkel AI raised $350 million at a $3.5 billion valuation, Reuters reported on September 22. The company supplies complex training data and simulated environments, with human specialists helping design and evaluate the work. Its CEO attributed growth to demand for more demanding AI training tasks.
My take: This is a useful reminder for anyone studying analytics or working in AI evaluation: data quality and human judgment remain central to the business. The flashy model gets the stage. Someone still has to check its homework.
6. Businesses are asking what a usable AI answer actually costs
A September 24 recap published by Reuters Events highlighted a practical measure discussed at its Austin conference: “cost per accepted output.” The idea includes human review, error handling, and fallback work when calculating what an AI result really costs. This source is the organizer’s conference recap, rather than an independent audit of the companies presenting.
My take: That is a more useful question than how quickly a system can produce 40 paragraphs. If an employee spends an hour repairing the result, the productivity calculation needs that hour included. A faster first draft is only helpful if it gets you to a better finished product sooner.
Read the Reuters Events recap on GlobeNewswire.
7. College football meets AI research at Notre Dame
Notre Dame’s Data, AI, and Computing Initiative is running a 2026 event series called Football Fr(AI)days. Its schedule includes “Touchdowns in Science with AI” on October 9 and “The Future of Human-Centered AI” on November 6. The series spotlights university research and discussions about AI’s role in society.
My take: This is a campus research connection to football season, not evidence that an AI system is calling Notre Dame’s plays. Still, pairing football weekends with accessible research discussions is a clever way to get people into the conversation. Come for the touchdowns. Stay for the questions about your future robot coworkers.
See Notre Dame’s official event series.
8. Knoxville watch: Trump is expected at Tennessee–Texas
WVLT reporting carried by WSMV says Trump plans to attend Tennessee’s game against Texas at Neyland Stadium on Saturday, September 26, citing the White House press office. Knoxville police said fans should expect increased security.
This is a local football and presidential-visit story. The reporting does not establish an AI announcement connected to the Knoxville trip. The separate September 29 AI meeting is the technology-policy item to watch.
My take: Neyland already knows how to generate a national spectacle. Adding a presidential visit should make the weekend even busier. If you’re heading downtown, check current university and police updates before leaving. No artificial intelligence is required to predict that parking will test your character.
Read the WVLT report via WSMV.
Reporting and scheduled events reflect sources available on September 25, 2026. Commentary is labeled separately. Follow The Drone Report for more AI headlines.
Leave a Reply