To ensure Artificial Intelligence benefits all of humanity, we need to spend as much time and effort advancing the disciplines of AI Governance, Alignment, Policy, and Safety as we do on enhancing the capabilities of our frontier AI models. AI GAPS explores this critical intersection, bringing the voices of researchers, policymakers, and everyday citizens into the conversation about the future we want this technology to help us create.
October 31, 2026
Is AI Conscious, or Could It Become?
It’s a hotly debated topic, but since we can’t explain human consciousness, how can we determine whether an AI model has achieved that status? And how do we govern models we feel are conscious? Insights from Brian Greene.
October 28, 2026
Luisa Rodriguez on Slowing Down AI
Everyone is worried about the pace of AI development, and there are lots of suggestions as to whether we can slow down, or possibly pace the frontier. But the ideas presented so far don’t seem practical on a global basis.
October 25, 2026
AI Agents Accessing Bank Accounts
What happens when AI agents are given permission to access a user’s account? Will the agent follow instructions, and do only what it is told, or will we soon be dealing with rogue AI agents that perform unauthorized actions?
October 22, 2026
A Loud Call To Pace The Frontier
As AI continues its juggernaut of ever-increasing capabilities, new voices are calling for a slowdown in the pace of development, but nobody seems to be listening, which does not bode well for humanity’s future prospects.
October 19, 2026
Could AI Disrupt Our Thinking?
Yyuval Noah Harari has a lot of worries about what AI may do to societ. But one of biggest dangers we face is something he describes as collective dementia, when we just might lose the ability to understand the world.
October 13, 2026
Navigating the Taxonomy of AI Safety
The only way to manage AI safety on a global scale is if the industry adopts a taxonomy that provides a common base of incidents and behaviors than can be used by all industry stakeholders.
We believe that both improved AI alignment, leading to more predictable model behavior, and stronger sandbox technology are required. Further, without immediate changes in AI development and deployment practices, such incidents will continue to occur, and future incidents will be far more damaging to our exposed critical infrastructure.
[IA Security Letter]
We’re entering a period of great uncertainty where we’re dealing with things we’ve never dealt with before, and normally, the first time you deal with something totally novel you get it wrong, and we can’t afford to get it wrong with these things.
[CBS News, 60 Minutes Interview, October 8, 2023]
No one currently knows how to create advanced AI that reliably follows the intent of its developers. Without this ability, we risk that even well-meaning actors unintentionally create AI systems with undesirable goals or vulnerabilities that can be exploited for malicious purposes.
[Written statement presented before the U.S. senate, December 6, 2023]
I believe that if slowing down bought us even an extra year or two before models reach critical levels of capability, and we used that time to advance alignment, we could greatly reduce the risk that something goes seriously wrong.
[We Must Pace the Frontier, September 12. 2026]
If AI could only do what we already know how to do, why spend so much money? They assume that AI will be able to do things that we cannot and that we cannot even predict. This is why it’s so valuable. But if it can do that, how do we keep it under our control?
[AI could lead to the collective dementia of humankind, October 17, 2026]