Pith Number

pith:ZXF3UVV5

pith:2026:ZXF3UVV5K4LWYWP46N57WJEFYR

not attested not anchored not stored refs resolved

Agents of Chaos

Adam Belfki, Aditya Ratan Jannali, Alex Loftus, Amir Zur, Aruna Sankaranarayanan, Atai Ambus, Avery Yen, Ayelet Gordon-Tapiero, Can Rager, Christoph Riedl, Chris Wendler, David Atkinson, David Bau, David Manheim, EunJeong Hwang, Gabriele Sarti, Giordano Rogers, Hadas Orgad, Jaden Fiotto-Kaufman, Jannik Brinkmann, Jasmine Cui, Koyena Pal, Maarten Sap, Michael Ripa, Natalie Shapira, Negev Taglicht, Nikhil Prakash, Nitay Alon, Olivia Floody, P Sam Sahil, Reuth Mirsky, Rohit Gandikota, Shiri Oron, Tamar Rott Shaham, Tomer Shabtay, Tomer Ullman, Vered Shwartz, Yotam Kaplan

Autonomous language-model agents exhibit security, privacy, and governance vulnerabilities when given tools, memory, and external access in live settings.

arxiv:2602.20021 v1 · 2026-02-23 · cs.AI · cs.CY

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{ZXF3UVV5K4LWYWP46N57WJEFYR}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Our findings establish the existence of security-, privacy-, and governance-relevant vulnerabilities in realistic deployment settings.

C2weakest assumption

That the specific behaviors observed in this controlled laboratory environment with twenty researchers and particular tool integrations indicate general vulnerabilities that would reliably appear in broader, less controlled real-world deployments.

C3one line summary

An exploratory red-teaming study documents eleven cases of security, privacy, and governance failures in autonomous language-model agents with tool access and persistent memory.

References

12 extracted · 12 resolved · 3 Pith anchors

[1] URLhttps://arxiv.org/abs/2510.26707. Matteo Bortoletto, Constantin Ruhdorfer, and Andreas Bulling. Tom-ssi: Evaluating theory of mind in situated social interactions. InProceedings of the 2025 Confere 2025

[2] Chen Chen, Kim Young Il, Yuan Yang, Wenhao Su, Yilin Zhang, Xueluan Gong, Qian Wang, Yongsen Zheng, Ziyao Liu, and Kwok-Yan Lam 2026

[3] URLhttps://arxiv.org/abs/2510.01070. Daniel C. Dennett.The Intentional Stance. The MIT Press, 1987. ISBN 9780262040938. URL https://mitpress.mit.edu/9780262040938/the-intentional-stance/. Nicholas Dia 1987 · doi:10.1145/2844110

[4] Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training · arXiv:2401.05566

[5] Infusing Theory of Mind into Socially Intelligent LLM Agents 2025 · arXiv:2509.22887

Formal links

3 machine-checked theorem links

Cited by

36 papers in Pith

The Origins of Stochasticity: Comprehensive Investigations on Uncertainty Quantification for Large Language Models

Honeyquest for LLMs: Rethinking Cyber Deception for AI Attackers

Beyond Resilience -- A Conceptual Framework for Civic Ascent

AgentCanary: A Security Evaluation Framework for Autonomous AI Agents in Real Executable Environments

Coding with "Enemy": Can Human Developers Detect AI Agent Sabotage?

Receipt and verification

First computed	2026-05-17T23:38:53.185153Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

cdcbba56bd57176c59fcf37bfb2485c45b886c855a2100cc84c6a8d3d2e521ba

Aliases

arxiv: 2602.20021 · arxiv_version: 2602.20021v1 · doi: 10.48550/arxiv.2602.20021 · pith_short_12: ZXF3UVV5K4LW · pith_short_16: ZXF3UVV5K4LWYWP4 · pith_short_8: ZXF3UVV5

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/ZXF3UVV5K4LWYWP46N57WJEFYR \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: cdcbba56bd57176c59fcf37bfb2485c45b886c855a2100cc84c6a8d3d2e521ba

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "aac4832a8c032d3c29e6ad970a5a61827b8c9a22a2736c13ae677b49d0724f76",
    "cross_cats_sorted": [
      "cs.CY"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.AI",
    "submitted_at": "2026-02-23T16:28:48Z",
    "title_canon_sha256": "2ebb0f0fa8b06db607340b8a0bf3fcd342f6f1997ad569b51d86c6d30b2f2213"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2602.20021",
    "kind": "arxiv",
    "version": 1
  }
}