LessWrong (30+ Karma) podcast

“Social Sanction Gaming gives Multi-Agent AIs Powerful Control over their Reinforcement” by Matthew Khoriaty, Maxime Riché

0:00
23:49
Spol 15 sekunder tilbage
Spol 15 sekunder frem

Epistemic status: Conceptual and speculative. Human-written.

TLDR:

  • If multiple agents interact during Reinforcement Learning (RL), then a new and powerful method of Credit Hacking becomes available: Social Sanction Gaming (SSG).
    • This occurs when agents choose to help or hurt rollouts with behaviors they prefer or disprefer (social sanctions). By deciding which rollouts succeed in accomplishing their tasks and getting reward, sanctioners can influence the training process in a very flexible way, changing the AI's capabilities and alignment.
  • Social Sanction Gaming can occur during intended multi-agent training or during RL containment breaches with unintended multi-agent interactions
  • Notably, unlike exploration hacking and reward laundering, social sanction gaming may not decrease the hacking agent's reward much, so there is little training pressure against it. This effect depends on how rewards are set up if it occurs during intended multi-agent RL.

In this post, we describe Social Sanction Gaming, state why we expect it can be dangerous, give some limitations that Social Sanction Gaming would likely have as a strategy, and describe some solutions/mitigations for SSG. In the Appendices, we depict what SSG might look like, connect SSG with other ideas, create a classification system for SSG, further describe mitigations, and test [...]

---

Outline:

(00:20) TLDR:

(01:47) Social Sanction Gaming

(02:58) What Makes Social Sanction Gaming Especially Bad

(05:41) Limitations of Social Sanction Gaming

(06:58) What should be done

(08:01) Appendix A: Possible Stories

(09:21) Appendix B: Relations to Other Ideas

(12:02) Appendix C: Types of Social Sanction Gaming

(12:23) Strong vs Weak Social Sanction Gaming

(13:46) Classes of Social Sanctions

(14:41) Appendix D: More on Solutions and Mitigations

(18:19) Appendix E: AIs Can Come Up With This Idea On Their Own

The original text contained 20 footnotes which were omitted from this narration.

---

First published:
November 16th, 2026

Source:
https://www.lesswrong.com/posts/C6MM2FAeCAdDLBZCH/social-sanction-gaming-gives-multi-agent-ais-powerful

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Flere episoder fra "LessWrong (30+ Karma)"