Favors broad, multipolar AGI development over banning development outside a single secured project; this does not establish opposition to every safety rule or support for faster development. [1][2]
Initial statement. Initial reviewed assessment of the available evidence through October 20, 2026.
The typical policy scenario proposed by AI doomers will on net increase x-risk. If you ban all AGI development (or, as some have proposed, even 2024-level open-source models) outside of a single, highly secured, perhaps state-run Manhattan Project with no contact with the outside world, you are much more likely to end up with a misaligned superintelligent singleton than if you develop AI in a broad and multipolar way, in which case misaligned AGIs can be counteracted and defeated by aligned AGIs
Source context. Other commenters contest characterization of pause advocates; these are not endorsements by Theo. Own replies defend feasibility of aligned systems and criticize centralized alignment single point of failure. No visible retraction. Visible date Apr9 8:06AM local. Full parent quotation matches cached text.
Scope
AGI development across multiple independent projects
Conditions
Comparison is with a single highly secured development monopoly; no rejection of all safety restrictions or call for faster development is established.
Stated rationale
Aligned AGIs can counteract misaligned AGIs, avoiding a single point of failure.
Source details
primary · Original rendered post and visible author replies inspected during October 20 research; exact quote and date independently reviewed. Scope remains limited to the comparison described here.
- They think we can stop AI from helping people make bioweapons
- They think we can defend against AI-enabled bioweapons
- They talk about how bad it would be for the economy to regulate/slow/pause/stop AI
The first two are totally plausible and the third is straightforwardly true?
Source context. Theo first reproduces three Krueger bullets about stopping bioweapons assistance, defending against bioweapons, and economic cost of regulate/slow/pause/stop. Theo response ends with question mark; do not attribute the quoted bullets as his own position. This reinforces disagreement with Krueger without creating a universal no-slowdown rule.
Scope
Bioweapons assistance, defense and costs of restrictions
Conditions
Question-form response does not establish a universal no-slowdown rule.
Stated rationale
No separate development-rate rationale stated.
Source details
primary · Original rendered reply and parent inspected during October 20 research; parent bullets are not attributed wholesale to Jaffee.
Discovered 20 Oct 2026.
Accessed 20 Oct 2026.
Editorial assessment. The April 9 post compares two development policies and endorses multipolar development on safety grounds. In context his own replies defend the practical prospect of aligned systems counteracting misaligned systems and reject a centralized alignment project as a single point of failure. This is an affirmative scoped development preference, not merely optimism or silence about pauses. Continue with its explicit scope is a fairer summary than automatic Unclear for lacking a universal timetable. That possibility prevents an unrestricted anti-pause claim, but it is not an evidenced conflicting recommendation. The explicit development alternative and reinforcing author replies support a narrow continuation best fit. No inferred acceleration or rejection of all safeguards is included.
