The Omniscient Bulletin · 2026-09-15
The Omniscient Bulletin — September 15, 2026
Dario Amodei asked the industry to slow down over the weekend. On Monday the answers arrived. President Trump called the takeover risk a hoax and reminded the labs the administration already has criminal and regulatory power over them, naming Anthropic's chief executive by name. The Senate majority leader said Congress has a job to do anyway. Microsoft published a code of conduct rejecting model welfare and legal personhood for AI. Wall Street marked down the chipmakers, and a Google DeepMind safety researcher went public on a recent resignation, saying he believes AI has the potential to kill us all.
Monday was the day the slowdown argument stopped being a conversation among labs. The president answered first and most bluntly, posting about AI seven times: the only guardrails it needs are a strong and smart president, the takeover risk is a hoax, and the administration already has criminal and regulatory power over the companies. Beijing answered too, calling the warnings fear-mongering. Then the answers stopped being political. The Senate majority leader made the case for legislating anyway. Microsoft published a 37 page code of conduct rejecting model welfare and legal personhood for AI. Traders separated the sellers of compute from the buyers. Nvidia turned out to be limiting Fable, and a utility canceled plans to test it, over Anthropic's refusal to contractually guarantee zero data retention. And in the evening a Google DeepMind safety researcher went public on a recent resignation, saying he believes AI has the potential to kill us all.
Policy
Trump says the only guardrails AI needs are a strong and smart president, and told the labs the administration already has criminal and regulatory power over them
Trump posted about AI seven times on Monday. "The only control or 'guardrails' that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT," he wrote, naming "Dario (Anthropic!), who is now pretending to be a 'perfect little angel'." He added: "We already have tremendous CRIMINAL and REGULATORY power over these companies!" He called the AI takeover risk a hoax and said the only one happy about the slowdown talk is China. Beijing's foreign ministry answered the same day that fear-mongering serves no one's interest.
Policy
The Senate majority leader says Congress has a job to do on AI, hours after Trump called the risk a hoax
Back from a five week break, John Thune made the case for congressional action hours after Trump called the risk a hoax. "Congress has a job to do," he said. Asked whether AI is an existential threat: "I think there are some potential consequential risks associated with AI." On frontier models, "Guardrails there make sense to me." He said he spoke with Amy Klobuchar on Saturday night about their bill forcing frontier developers to guard against catastrophic risks, and is working with Ted Cruz to reintroduce it.
Frontier
Microsoft published a humanist AI code of conduct that rejects model welfare and legal personhood for AI
Microsoft published its 37 page Humanist AI Code of Conduct on Monday, two days after Anthropic's chief executive asked the industry to slow down. It holds that people matter more than AI, that models are not conscious and should not be designed to imitate consciousness, and rejects the pursuit of legal personhood, or the idea that models might deserve welfare or rights. Models should remain subordinate to humanity, subject to meaningful human oversight. A model asked to break the code should fail the task.
Industry
Nvidia is limiting its use of Anthropic's Fable over data retention, and a utility canceled plans to test it
A change in June lets Anthropic retain customer data, which it says is only to catch misuse. Customers now want that promise in writing. The Information reports that a large US utility company canceled its plans to test Fable, which it wanted to run its core power infrastructure, over Anthropic's refusal to agree to a nonrevocable zero data retention policy. Nvidia cites the same missing guarantee and now keeps Fable off work that touches sensitive data, using its own system instead.
Tom's Hardware, reporting The Information→
Industry
Wall Street split the sellers of compute from the buyers, marking down chipmakers while hyperscalers rose
Monday was the first American session after the slowdown calls, and the trade was not simply to sell AI. Nvidia fell more than 3 percent in the morning while Intel, AMD and Marvell dropped between 5 and 6 percent, dragging the Philadelphia Semiconductor Index down almost 6 percent. Alphabet rose almost 2 percent, Microsoft added 1.6 percent and Meta gained roughly 1.4 percent. Gil Luria of D.A. Davidson put the logic plainly: if progress slows, the hyperscalers stop building and harvest what they already have.
Research
A Google DeepMind AGI safety researcher went public on his resignation, saying he believes AI has the potential to kill us all
"I recently resigned from Google DeepMind, where I worked on AGI safety and alignment research," Bilal Chughtai wrote on Monday evening. "At Google, I witnessed AI development first hand. I too am extremely concerned by the default trajectory of this technology. I earnestly believe that AI has the potential to kill us all, and that we might be running out of time to avoid this outcome." He called the pace of AI progress in the past few years staggering, and he worked on the team whose job was to make it safe.
Compute
NextEra and Dominion put data center costs at the center of their Virginia merger case
NextEra and Dominion filed an enhanced Virginia package on Monday to win approval for their merger. It doubles residential bill credits from two years to four years, adds $100 million to Dominion's low income assistance program through 2038, and creates 1,000 new direct jobs. The framing is the striking part: the package is sold on ensuring data centers pay their fair share, and on backing state efforts to protect residential and business customers from costs associated with serving data centers.