CVE-2025-14009

The CVE describes a critical remote code execution flaw in the nltk/nltk project, via the downloader component. The vulnerable code path is in _unzip_iter within nltk/downloader.py, which calls zipfile.extractall() without path validation or security checks, allowing a malicious zip package to ex...

CVE-2019-14751

CVE-2019-14751 affects the Python package NLTK (downloader) prior to 3.4.5. The vulnerability is a directory traversal (Zip Slip) during extraction of an NLTK ZIP package, enabling an attacker to write arbitrary files. OpenSUSE and Fedora advisories indicate the fix is to upgrade to 3.4.5 or late...

CVE-2026-54293

CVE-2026-54293 affects NLTK’s nltk.data.load() in Python. A TOCTOU-style flaw lets an attacker bypass the unsafe-path regex (UNSAFE_NO_PROTOCOL_RE) by using URL-encoded path separators (e.g., %2f, %2e%2e) and then decoding, enabling arbitrary local file reads prior to the fix. Affected until vers...

CVE-2026-0847

NLTK up to 3.9.2 contains a path traversal vulnerability in CorpusReader classes (WordListCorpusReader, TaggedCorpusReader, BracketParseCorpusReader) that can lead to arbitrary file reads on the server. Root cause is improper sanitization/validation of file paths, enabling access to sensitive fil...

CVE-2021-3828

CVE-2021-3828 affects the Python nltk package with a vulnerability to ReDoS (inefficient regular-expression handling). Connected sources show fixes in newer nltk releases (e.g., updates to 3.6.x/3.7, and OpenSUSE/SUSE advisories mentioning python-nltk fixes) and advisories noting the issue in Ubu...

CVE-2021-3842

CVE-2021-3842 affects the Python NLTK package; the vulnerability is an Inefficient Regular Expression Complexity (REDoS) in certain RegexpTaggers, leading to potential denial of service. Exposed components: nltk/nltk (regex processing). Public sources assign a high severity (CVSS v3.1 base 7.5; i...

CVE-2021-43854

CVE-2021-43854 affects the Python package NLTK. Versions prior to 3.6.5 are vulnerable to a Regular Expression Denial of Service (ReDoS) in the affected tokenizers: PunktSentenceTokenizer, sent_tokenize, and word_tokenize. The issue arises from inefficient regex backtracking on crafted inputs, ca...

CVE-2026-12372

A Server-Side Request Forgery (SSRF) exists in nltk/nltk versions 3.9.4 and the current develop branch. The function nltk.pathsec.validate_network_url() fails to reject IPs in RFC 6598 shared address space (100.64.0.0/10) because Python’s ipaddress classification does not mark such addresses as i...

CVE-2026-33231

CVE-2026-33231 affects NLTK’s nltk.app.wordnet_app in versions 3.9.3 and earlier, where the WordNet Browser HTTP server started in default mode can be remotely shutdown by an unauthenticated GET request to /SHUTDOWN%20THE%20SERVER, causing a denial of service via os._exit(0). IBM’s bulletin/Secur...

CVE-2026-0846

The CVE concerns nltk 3.9.2, specifically the filestring() function in nltk.util, which opens user-supplied file paths without proper sanitization. This allows arbitrary file read by passing absolute or traversal paths, enabling access to sensitive system files. Exploitation can occur locally or ...

CVE-2026-0848

NLTK versions ≤3.9.2 are vulnerable due to the StanfordSegmenter loading external Java .jar files without verification or sandboxing. An attacker can supply or replace the JAR, enabling arbitrary Java bytecode execution at import time via unvalidated classpath input. Potential attack vectors incl...

CVE-2026-12252

CVE-2026-12252 affects nltk/nltk versions 3.9.3 and earlier, leaving five Stanford interface classes (StanfordPOSTagger, StanfordNERTagger, StanfordParser, StanfordDependencyParser, StanfordNeuralDependencyParser) vulnerable to untrusted JAR code execution. These classes accept user-controlled JA...

CVE-2026-12261

The CVE concerns nltk.downloader in nltk/nltk ≤ 3.9.4. The downloader extracts package archives into shared namespaces (corpora/ and taggers/) instead of package-isolated roots and validates integrity only after extraction. This design flaw enables one package to overwrite another package’s trust...

CVE-2026-33236

CVE-2026-33236 affects the NLTK downloader in versions up to 3.9.3, where remote XML index processing does not validate the subdir and id attributes. This allows an attacker-controlled XML index server to supply path traversal values (e.g., ../) that can lead to arbitrary directory creation, file...

CVE-2026-33230

NLTK WordNet Browser (nltk.app.wordnet_app) in versions <= 3.9.3 contains a reflected XSS in the lookup_ route, where attacker-controlled word data is echoed into HTML without escaping, enabling arbitrary HTML/JS execution in the local browser context. Public advisories and vendor notices conf...

CVE-2025-71408

The CVE-2025-71408 issue affects NLTK up to version 3.9.2 (pre-3.9.3). The vulnerability lies in nltk.collocations’ main path, where command‑line arguments are passed directly to eval() to form BigramAssocMeasures suffixes without a allowlist or sanitization, enabling an attacker who can control ...

CVE-2026-79657

NLTK (Python Natural Language Toolkit) versions before 3.10.3 are affected by a remote code execution vulnerability in its allowlisted pickle deserialization mechanism. The root cause is that the safe pickle loaders trust entire module namespaces rather than restricting to specific safe callables...

CVE-2026-62388

NLTK (Python Natural Language Processing toolkit) versions before 3.10.0 ship with an insecure default in pathsec.py : the ENFORCE flag is set to False, causing all security validation functions to emit warnings rather than raising exceptions. This means path traversal and pickle deserialization ...

CVE-2026-79675

NLTK before 3.10.3 fails to validate JVM options passed through the per-call options parameter in the java() function. Attackers can inject dangerous JVM flags such as -agentpath , -javaagent , or @argfile into Stanford wrapper classes, achieving arbitrary code execution . The vulnerability is ra...

CVE-2026-66393

NLTK versions before 3.9.4 are affected by an unbounded recursion vulnerability in JSONTaggedDecoder.decode_obj() . An attacker can supply a deeply nested JSON structure that exceeds Python's recursion limit, triggering an unhandled RecursionError that crashes the process, resulting in denial of ...

CVE-2026-63310

NLTK (Natural Language Toolkit) versions before 3.9.3 contain a missing post-download integrity verification flaw in the downloader module . After downloading packages, the module fails to validate file integrity before extraction. An attacker positioned for a man-in-the-middle attack or performi...

CVE-2026-62383

nltk versions before 3.10.2 contain a symlink-based arbitrary file read vulnerability in the IPIPANCorpusReader component. The root cause is a complete bypass of nltk.pathsec validation: an attacker with local access can place a symlink in the corpus root directory and invoke channels(), domains(...

CVE-2026-71514

NLTK 3.9.4 through 3.10.2 contains a path traversal vulnerability in CrubadanCorpusReader._load_lang_ngrams . The function joins the corpus root with a value read from the corpus table.txt mapping file and opens the result using the builtin open() instead of the pathsec -validated opener. Because...

CVE-2026-78681

NLTK versions before 3.10.3 are vulnerable to a denial of service via XML entity expansion. The library uses xml.etree.ElementTree to parse XML in multiple modules, and this parser honors entity declarations in document DTDs. An attacker can supply a crafted XML payload containing nested entity d...

CVE-2026-78683

NLTK versions ≤3.9.4 contain an unsafe pickle deserialization vulnerability in the TransitionParser.parse() method (nltk/parse/transitionparser.py). The method calls pickle_load() with restricted=False, routing through WarningUnpickler , which does not override find_class() and thus permits arbit...

CVE-2026-80205

NLTK versions before 3.10.0 are vulnerable to a ReDoS (Regular Expression Denial of Service) in the Text.findall() and TokenSearcher.findall() methods. These methods accept user-supplied regular expressions without validation or a timeout mechanism, allowing an attacker to supply crafted patterns...

CVE-2026-63312

NLTK before 3.10.0 contains an arbitrary local file read vulnerability in StreamBackedCorpusView . The root cause is that the component calls builtins.open() directly instead of the sandboxed pathsec.open() , effectively bypassing the pathsec.ENFORCE security mechanism. An attacker who controls t...

CVE-2026-63311

NLTK (Python NLP library) versions

CVE-2026-78682

NLTK before 3.10.3 is affected by a server-side request forgery (SSRF) vulnerability in nltk.pathsec.urlopen and its callers (nltk.data.load, nltk.downloader.Downloader.index/download). The root cause is that when an HTTP proxy is configured, proxy-handler inheritance disables the safe HTTP/HTTPS...

CVE-2026-81726

NLTK through 3.10.3 contains a path traversal vulnerability in its model-artifact APIs. The root cause is that TransitionParser, AveragedPerceptron, PerceptronTagger, and maxent parameter APIs bypass pathsec enforcement by performing raw file operations on caller-controlled paths, allowing attack...

CVE-2026-62385

NLTK versions before 3.10.0 contain a path traversal vulnerability in FramenetCorpusReader and NKJPCorpusReader . By supplying unsafe selectors or poisoned index state, an attacker can exploit the frame_by_name, doc, lu, and header methods to parse XML files outside the corpus root, enabling arbi...

CVE-2026-65915

NLTK versions before 3.10.0 contain a logic bug in FileSystemPathPointer.open() where the sandbox validation check compares a normalized path against itself, rendering the security check permanently inert. An attacker with low privileges can pass file:// URLs to nltk.data.load() to read arbitrary...

CVE-2026-70626

NLTK versions before 3.9.4 are affected by a symlink escape vulnerability in CorpusReader.open() . The root cause is that path validation is purely lexical and does not resolve symlinks, allowing a local attacker to place a symlink inside the corpus root to read arbitrary files outside the intend...

CVE-2026-62384

NLTK (Python Natural Language Toolkit) versions before 3.10.2 contain a symlink-based sandbox bypass in FramenetCorpusReader . An attacker who can place symlinks with names containing no path separators inside a corpus subdirectory can bypass the path validation guard and read arbitrary XML files...

CVE-2026-79676

NLTK versions before 3.10.3 are affected by a path traversal vulnerability in its corpus readers. The root cause is that these readers reopen root-derived paths using the built-in open() instead of nltk.pathsec.open(), which fails to resolve symlinks confined to a trusted root. An attacker who ca...

CVE-2026-80206

NLTK (a Python natural language processing library) versions before 3.10.3 contain a ReDoS vulnerability in the tgrep module. The _tgrep_node_action function compiles user-supplied regular expressions embedded in /regex/ pattern nodes and executes them via re.search against tree node labels witho...

CVE-2026-81724

NLTK before 3.10.3 contains an uncontrolled recursion vulnerability in nltk.featstruct.FeatStructReader . Unauthenticated, remote attackers can supply deeply nested feature-structure input—trivially crafted with nested brackets—that exceeds Python's recursion limit and raises an unhandled Recursi...

CVE-2026-79674

NLTK (Natural Language Toolkit) versions before 3.10.3 contain a path sandbox bypass vulnerability in corpus-reader constructors. Attackers can supply arbitrary corpus root paths to LinThesaurusCorpusReader and PanLexLiteCorpusReader , allowing them to read filesystem content and SQLite databases...

CVE-2026-81725

NLTK before 3.10.3 contains a Regular Expression Denial of Service vulnerability in the Pl196xCorpusReader component. The root cause lies in lazy regex patterns within the read_block method, which can be triggered by supplying malformed TEI blocks containing many unmatched opening tags. Attackers...

CVE-2026-81727

NLTK versions before 3.10.3 are affected by a filesystem containment bypass in the Downloader.download and Downloader.incr_download methods. An attacker with write access to a shared downloader directory can create hardlinks pointing to files outside the install root, which are then overwritten d...

CVE-2026-81722

nltk PorterStemmer in versions ≤ 3.10.2 is vulnerable to an inefficient-algorithmic-complexity denial of service in PorterStemmer.stem(). The root cause is that the _is_consonant() helper walks backward over the entire run of trailing 'y' characters on every call, and _measure() invokes it for ea...