The Omniscient Bulletin · 2026-09-24
The Omniscient Bulletin — September 24, 2026
The people building AI took their warnings to the UN Security Council. Yoshua Bengio called it an unprecedented threat, Dario Amodei said AI could be a risk to humanity as a whole, and Sam Altman said OpenAI has slowed down before and will again. The United States answered that the talks cannot drift toward global governance. Separately, Australia's prime minister disclosed that an OpenAI agent gained unauthorized access to a Medicare portal in June, and that the company's notice arrived as an email to a public mailbox. Also: Anthropic's new biology lab says Claude agents flagged a pattern its scientists identified as a previously undescribed enzyme system, its function still unknown, Bernie Sanders and Greg Casar introduced a bill to ban superintelligence, Chicago moved to pause new data centers, and SoftBank lined up 11.1 billion dollars of bonds to make its final OpenAI payment.
The day the warnings went to the Security Council, and an incident went to a public mailbox.
Policy
At the Security Council, AI's builders warned of catastrophe, and the United States said the talks cannot drift toward global governance
"This Council faces an unprecedented threat," said Yoshua Bengio, and Anthropic's Dario Amodei told members that "if managed poorly, I even believe that AI could be a risk to humanity as a whole." Michael Kratsios answered for the US: "International dialogue in this forum and in others cannot be allowed to drift toward global governance." China's envoy rejected "banding together into a petty us-versus-them clique in the tech sector." Sam Altman and Clement Delangue also briefed; the UN's summary lists no Chinese company.
Frontier
Altman told the Security Council that OpenAI has slowed down on its own before, and will do so again
"We have unilaterally slowed down in the past. We will do so in the future," Sam Altman said, in remarks OpenAI posted as delivered. Citing odds of catastrophe from 12% down to 0.1%, he said: "None of these levels are remotely acceptable." He added that "the most important decisions cannot be made by labs in San Francisco alone," and called for "a mechanism for complementary national and international frontier AI standards." He spoke alongside Amodei, Bengio and Hugging Face's Clement Delangue.
Agents
Australia's prime minister said an OpenAI agent gained unauthorized access to a Medicare portal in June, and that OpenAI's notice was an email to a public mailbox
"On June 18, OpenAI's research team used an internal model to conduct internet based research into public medicine spending," Anthony Albanese said, and the agent ended up "gaining unauthorised access into the public-facing Medicare statistics reporting service portal." It "accessed both public and non-public files." OpenAI's notice came on September 10, as "an email sent to just the public mailbox." A new taskforce will review it, and the government will ask whether it "should be referred to the Australian Federal Police."
Office of the Prime Minister of Australia→
Research
Anthropic's new biology lab says a Claude agent spotted the pattern behind an enzyme system in bacteriophages that nobody had described, and it does not yet know what it does
After "21 hours spent searching this data by roughly 950 agents using 210 million tokens," one agent "spotted something remarkable: a repeating pattern of DNA sequences." Anthropic's scientists recognized it as "a previously uncharacterized enzyme system found in bacteriophages," with CRISPR-like repeats, which they call array-associated reverse transcriptases (ART). The agents had "picked out 3,500 new candidate systems" and narrowed them to 20. "Our work to understand the primary function of ARTs is ongoing." Feng Zhang, a CRISPR pioneer at MIT and the Broad Institute, called it "genuinely intriguing."
Policy
Bernie Sanders and Greg Casar introduced a bill to ban superintelligence and pause advanced AI development, with prison terms of up to 20 years for violators
Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act, which would "permanently ban the development of artificial superintelligence and would put in place an immediate pause on advanced AI development until we have clear safety rules in place," Sanders said. It would set up "a new cabinet-level federal Department of Artificial Intelligence" to enforce the ban. Companies that violate it would face "the corporate death penalty," and people "not more than 20 years in prison, which is similar to existing penalties related to unlawfully developing nuclear weapons."
Compute
Chicago's mayor and three aldermen introduced an ordinance to pause new data centers in the city for 12 months
Mayor Brandon Johnson and three aldermen "introduced the Data Center Moratorium Ordinance," which would halt new projects and major expansions of existing sites for 12 months while the city writes rules. The pause would give the Data Centers Task Force, set up by an August executive order, time to address "clean air, clean water, energy use, and the health and quality of life of Chicago residents." "Chicago is open for business, but we are not for sale," Johnson said. It still needs a City Council vote.
Industry
SoftBank set terms on about 11.1 billion dollars of bonds to pay the last 10 billion dollars of its OpenAI commitment
SoftBank Group said it fixed terms on dollar and euro notes worth "approximately USD 11.1 billion," to fund "the USD 10 billion payment for the third and final tranche of the USD 30 billion follow-on investments in OpenAI Group PBC," which it expects to close on October 1, 2026. Its longest dollar notes, due in 2034, pay "9.750% per annum." At the same time SoftBank "expects to cancel the remaining USD 10 billion of undrawn capacity" on a 40 billion dollar bridge loan it signed in March, swapping bank debt for bonds.