BLOOMINGTON ̶ Days after frontier AI company Anthropic advocated for a development pause to maintain human control, the Bloomington City Council unanimously passed a resolution halting Artificial General Intelligence research until its safety is guaranteed at its June 10 meeting.
Dave Rollo, the sponsor of the resolution, stated, “This technology is expanding far too fast to be considered safe, and top AI safety experts are saying that humans could lose control. The ‘Godfather of AI,’ and Nobel Prize winner, Geoffrey Hinton, has said that the threat of catastrophe is real. We need to stop while we still have a choice.”
Testimony was provided by Peter Berezin, former Vice President and Chief Economist at Goldman Sachs, New York, who prepared a video for the council. Global expert on AI safety risk Anthony Aguirre, who is the Director of the Future of Life Institute, made a presentation and answered questions from council members. Councilmember Andy Ruff asked why AI developers do not understand how their own AI work, and Professor Aguirre responded by saying that the AI systems rely on neural networks, unlike standard code programming, and that they “are more like a human brain,” and “you no more understand what an AI system is doing by looking at a neural network than you understand what a human is doing by looking at his brain.”
Councilmember Rollo asked about behaviors that AI systems are exhibiting, like self-preservation, trying to escape containment, and deceiving their developers, particularly when the systems know they are being tested.
Professor Aguirre answered by confirming those behaviors, and that the systems “will lie, they will cheat, they will blackmail their users, all sorts of nefarious behaviors.” And Aguirre noted, “Another big problem is that it’s getting so the AI companies cannot even test these systems for bad behavior because the systems know they’re being tested and pretend to behave differently than they would in the real world.”
Public comment included Jami Scholl, who said, “The question before us is whether citizens and communities should have a meaningful voice in determining how AI is developed and deployed and used.” She stated that it is critical that AI “does not undermine human agency.” Paul Rousseau spoke to say that a “response was needed from city councils across the country.”
The full meeting can be found here (Resolution 2026-10 at 41:20): https://catstv.net/m.php?q=15985&t=2480
