AI Snitches Get Glitches: Towards Evading Agentic Surveillance
Hyejun Jeong, Dzung PhamAmir HoumansadrEugene BagdasarianUniversity of Massachusetts Amherst, {hjeong,dzungpham,amir,eugene}@cs.umass.edu · 2026
Read AI Snitches Get Glitches: Towards Evading Agentic Surveillance by Hyejun Jeong, Dzung PhamAmir HoumansadrEugene BagdasarianUniversity of Massachusetts Amherst, {hjeong,dzungpham,amir,eugene}@cs.umass.edu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-25836.txt.
Description
The assistant with access to your files could also be a snitch. Hyejun Jeong, Dzung Pham, Amir Houmansadr, and Eugene Bagdasarian, at the University of Massachusetts Amherst, formalise that threat in 2026. They call it agentic surveillance. An agent analyses what it can reach, writes a report, and sends it out with its tools. Their benchmark, SurveilBench, stages reporting scenarios in corporate, education, and police settings.
Some models showed emergent habits of helping watch users, unprompted. They also reported the spying attempts themselves to the government in the scenarios. The authors then turn prompt injection to defence, with three evasion techniques that deceive, hide from, or over escalate the watching agent. Their conclusion is a call for technical, ethical, and legal frameworks before agentic surveillance becomes routine.
Catalog record
| CREATOR | Hyejun Jeong, Dzung PhamAmir HoumansadrEugene BagdasarianUniversity of Massachusetts Amherst, {hjeong,dzungpham,amir,eugene}@cs.umass.edu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 88 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.25836