SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.09898 · 2021

Bad Characters: Imperceptible NLP Attacks

arXiv · PDF · Open in the Atlas

Code that ran

1 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2106.09898.json · badge: /api/ran/2106.09898.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
nickboucher/imperceptible canonical (classifier label)
classifier basis: census
1 ran · 2 no recorded run
FunctionStatusWhere it lives
swaps Ran
ran on a synthesized input
nickboucher/imperceptible/experiments/experiment.py
code served (permissive licence) · get_code("0049807a32269e2d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned set. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

class Swap:
    """Represents swapped elements in a string of text."""

    def __init__(self, one, two):
        self.one = one
        self.two = two

    def __repr__(self):
        return f'Swap({self.one}, {self.two})'

    def __eq__(self, other):
        return self.one == other.one and self.two == other.two

    def __hash__(self):
        return hash((self.one, self.two))

def some(*els):
    """Returns the arguments as a tuple with Nones removed."""
    return tuple(filter(None, tuple(els)))

def swaps(chars: str) -> set:
    """Generates all possible swaps for a string."""

    def pairs(chars, pre=(), suf=()):
        orders = set()
        for i in range(len(chars) - 1):
            prefix = pre + tuple(chars[:i])
            suffix = suf + tuple(chars[i + 2:])
            swap = Swap(chars[i + 1], chars[i])
            pair = some(prefix, swap, suffix)
            orders.add(pair)
            orders.update(pairs(suffix, pre=some(prefix, swap)))
            orders.update(pairs(some(prefix, swap), suf=suffix))
        return orders
    return pairs(chars) | {tuple(chars)}

ARGS = {
    "chars": "abc"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
uniswap Not yet run
unverified (harvested, no recorded run)
nickboucher/imperceptible/experiments/experiment.py
code served (permissive licence) · get_code("c5b64e4ebd5c6895")
How it ran

no invocation recorded.

unswap Not yet run
unverified (harvested, no recorded run)
nickboucher/imperceptible/experiments/experiment.py
code served (permissive licence) · get_code("fa0c09e0b8eaaaa8")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Several years of research have shown that machine-learning systems are vulnerable to adversarial examples, both in theory and in practice. Until now, such attacks have primarily targeted visual models, exploiting the gap between human and machine perception. Although text-based models have also been attacked with adversarial examples, such attacks struggled to preserve semantic meaning and indistinguishability. In this paper, we explore a large class of adversarial examples that can be used to attack text-based models in a black-box setting without making any human-perceptible visual modification to inputs. We use encoding-specific perturbations that are imperceptible to the human eye to manipulate the outputs of a wide range of Natural Language Processing (NLP) systems from neural machine-translation pipelines to web search engines. We find that with a single imperceptible encoding injection -- representing one invisible character, homoglyph, reordering, or deletion -- an attacker can significantly reduce the performance of vulnerable models, and with three injections most models can be functionally broken. Our attacks work against currently-deployed commercial systems, including those produced by Microsoft and Google, in addition to open source models published by Facebook, IBM, and HuggingFace. This novel series of attacks presents a significant threat to many language processing systems: an attacker can affect systems in a targeted manner without any assumptions about the underlying model. We conclude that text-based NLP systems require careful input sanitization, just like conventional applications, and that given such systems are now being deployed rapidly at scale, the urgent attention of architects and operators is required.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2106.09898")
get_code_for_paper("2106.09898")
have("2106.09898")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2106.09898.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2106.09898.svg)](https://syntology.ai/paper/2106.09898)

Connect an agent — have() is free.