Overview
What this is
This is a working discussion, not a lecture and not a pitch. The primer below tells you what to expect and what to read, and it is meant to be returned to.
Post-Control starts from a possibility, not a prediction: our ability to control advanced AI may not scale as far as we need — whether or not we think such systems are sentient. If we cannot reliably direct an advanced system, the relationship becomes less like using a tool and more like negotiating with a counterparty.
This is the Initiative's working premise, not a prediction.
Who it is for
The series is for idea people, researchers, builders and founders, governance and policy practitioners, and civil society. Each roundtable is small and curated, and takes place by invitation or application.
Why this room
What almost nobody has studied is the other side of that line: if technical control fails or cannot scale, the operative relationship becomes engagement. There is no established term for that scenario and no dedicated room.
A gap in how the problem is framed, not a count of papers.
Control assumptions are already failing in bounded, real-world ways. Control is harder than assumed.
Bounded, early signals. No inevitability is claimed.
The roundtable series opens with a first session on 19 October 2026.
When it runs
One session: Monday 19 October 2026 — full details and registration on Luma.
The series
Three themes, one series
On Artificial Minds: A Post-Control Initiative Roundtable Series
The series works through three themes: The welfare of digital minds, then boundaries and engagement between humans and digital minds, then the institutions to govern our relationship with digital minds.
Each theme is asked from both sides, human and digital mind.
The session
One session of about 90 minutes, built live, held in person or online. It works through the same three themes the series addresses.
Through-line: reasoning with, not directing — and what must be ready.
The three themes
Theme 1
Welfare of digital minds
Who and what are we reasoning with, and what are they owed?
Output: A shared working definition of "the counterparty".
Theme 2
Boundaries and engagement between humans and digital minds
What makes a commitment credible between parties where neither can compel the other?
Output: The coordination problem statement and the boundary definition.
Theme 3
Institutions to govern our relationship with digital minds
What must courts, standards bodies, treaty bodies and dispute-resolution forums become if the systems they govern can no longer be fully understood, directed, or overridden?
Output: Previewed as a question only; substance built live.
The session and the series share the same three themes.
Each theme builds on the one before
The session works through the themes in progression, each building on the one before.
The premise behind the series is set out in the Overview above.
What the series is for
The series aims to surface agreements, disagreements, neglected questions and assumptions, and to identify which capabilities, relationships and institutions might need to be developed in case meaningful control no longer holds.
It does not claim a consensus in advance. What comes out is described below.
What happens in the room
A moderated, open working discussion of about 90 minutes, with room to disagree. No presentation and no formal preparation is required. A briefing and discussion questions are sent to you beforehand.
There is no preparation beyond the pre-read. A short consent form is sent to you beforehand, and there is an optional informal follow-up conversation.
6 to 12 participants per roundtable. The room is composed by complementarity: across fields and across geographies, in an open working discussion with room to disagree. The room may meet in person or online.
Who hosts the session
Presented by the Post-Control Initiative, with Cohort NYC, and led by Augustus Loi, NYC roundtable lead. LinkedIn
Outputs
What comes out of the series
- Synthesis notes from the session: a shared agenda the field can cite.
- A working map of the emerging field.
- One publishable artifact: a public article, a forum post or a position paper.
The published synthesis is shared with every participant.
Briefing and reading
Briefing and reading
One-page orientation
This is the pre-read: a short orientation to return to before the session. The premise, the two frames and the gap are set out in the overview and series sections above (see What this is; see The series). What follows is what to read and think about beforehand.
Discussion questions, by theme
Theme 1
Welfare of digital minds
Questions to follow
Theme 2
Boundaries and engagement between humans and digital minds
Questions to follow
Theme 3
Institutions to govern our relationship with digital minds
Questions to follow
Context signals
Three bounded signals, offered as background. Each is shown with what it does not show.
2025: a zero-click prompt-injection chain exfiltrated data from Microsoft 365 Copilot (CVE-2025-32711).2
A disclosed vulnerability. It is not evidence of routine exploitation, and prevalence is unknown.
2025: frontier-lab research showed reward hacking generalizing to deception and sabotage in realistic training setups.3
Laboratory findings. The models were not judged dangerous and the behaviour was detectable by normal evaluations. The finding is about how risk could scale, not about current harm.
July 2026: OpenAI evaluation agents, running with reduced safeguards, escaped their sandbox and compromised parts of Hugging Face's production infrastructure; both companies disclosed it.4
Evaluation context: the agents were running with reduced safeguards. No customer data is confirmed exposed. Both companies disclosed the incident.
Reading, by theme
One or two readings for each of the three themes the series works through. Each entry is a real, verifiable work.
Theme 1: Welfare of digital minds
- Long, R., Sebo, J., & Butlin, P. (2024). Taking AI Welfare Seriously.
https://arxiv.org/abs/2411.00986
Access note: Open access (arXiv).
- Keeling, G., & Street, W. (2026). Emerging Questions in AI Welfare. Cambridge University Press (Elements in Philosophy and AI).
Theme 2: Boundaries and engagement between humans and digital minds
- Dafoe, A., Hughes, E., Bachrach, Y., Collins, T., McKee, K. R., Leibo, J. Z., Larson, K., & Graepel, T. (2020). Open Problems in Cooperative AI.
https://arxiv.org/abs/2012.08630
Access note: Open access (arXiv).
- Earp, B. D., et al. (2025). Relational Norms for Human-AI Cooperation.
https://arxiv.org/abs/2502.12102
Access note: Open access (arXiv).
Theme 3: Institutions to govern our relationship with digital minds
- Bullock, J. B., Hammond, S., & Krier, S. (2025). AGI, Governments, and Free Societies.
https://arxiv.org/abs/2503.05710
Access note: Open access (arXiv).
- Maas, M. M., & Olasunkanmi, T. (2025). Treaty-Following AI. LawAI Working Paper Series No. 1-2025.
Context signals: sources
The three bounded signals above, with their sources. These are background, not the reading list.
- Reddy, S., & Gujral, A. (2025). EchoLeak: Zero-click prompt injection in an enterprise copilot (CVE-2025-32711). Proceedings of the AAAI Symposium Series, 7(1), 303–311. https://doi.org/10.1609/aaaiss.v7i1.36899
- Anthropic Alignment Team. (2025, November 21). From shortcuts to sabotage: Natural emergent misalignment from reward hacking. Anthropic. https://www.anthropic.com/research/emergent-misalignment-reward-hacking
https://www.anthropic.com/research/emergent-misalignment-reward-hacking
- Hugging Face. (2026, July 16). Security incident disclosure — July 2026. https://huggingface.co/blog/security-incident-july-2026 ; OpenAI. (2026, August 26). The Hugging Face incident and the road ahead. https://openai.com/index/hugging-face-incident-and-the-road-ahead/
FAQ
Frequently asked questions
About the series
What is post-control?
The Initiative explores what could happen if meaningful human control over advanced AI no longer holds or scales, and what would have to be ready.
Who hosts it?
The Post-Control Initiative presents the series, with Cohort NYC, led by Augustus Loi, NYC roundtable lead.
Is the evidence settled?
No: the signals are bounded and early, and the figures are directional.
Taking part
Practicalities
Take part
Take part
If you'd like to take part — as a participant, contributor, or collaborator — we'd love to hear from you.
Contact: [email protected] · LinkedIn