PT-2026-59891

Impact When SparseSparseMaximum is given invalid sparse tensors as inputs, it can give an NPE. python import tensorflow as tf tf.raw ops.SparseSparseMaximum a indices=1, a values = 0.1 , a shape = 2, b indices=, b values =2 , b shape = 2, Patches We have patched the issue in GitHub commit...

PT-2026-59802

Impact An input sparse matrix that is not a matrix with a shape with rank 0 will trigger a CHECK fail in tf.raw ops.SparseMatrixNNZ. python import tensorflow as tf tf.raw ops.SparseMatrixNNZsparse matrix= Patches We have patched the issue in GitHub commit f856d02e5322821aad155dad9b3acab1e9f5d693...

PT-2026-59844

Impact The implementation of Conv2DBackpropInput requires input sizes to be 4-dimensional. Otherwise, it gives a CHECK failure which can be used to trigger a denial of service attack: python import tensorflow as tf strides = 1, 1, 1, 1 padding = "SAME" use cudnn on gpu = True explicit paddings =...

PYSEC-2026-3171 TensorFlow vulnerable to OOB write in `scatter_nd` in TF Lite

Impact The ScatterNd function takes an input argument that determines the indices of of the output tensor. An input index greater than the output tensor or less than zero will either write content at the wrong index or trigger a crash. Patches We have patched the issue in GitHub commit...

PYSEC-2026-3221 Missing validation causes denial of service via `LoadAndRemapMatrix`

Impact The implementation of tf.rawops.LoadAndRemapMatrix does not fully validate the input arguments. This results in a CHECK-failure which can be used to trigger a denial of service attack: python import tensorflow as tf ckptpath = tf.constant...

Missing validation causes denial of service via `SparseTensorToCSRSparseMatrix`

ImpactThe implementation of tf.rawops.SparseTensorToCSRSparseMatrix does not fully validate the input arguments. This results in a CHECK-failure which can be used to trigger a denial of service attack:pythonimport tensorflow as tfindices = tf.constant53, shape=3, dtype=tf.int64values =...

PYSEC-2026-3202 Missing validation causes denial of service via `UnsortedSegmentJoin`

Impact The implementation of tf.rawops.UnsortedSegmentJoin does not fully validate the input arguments. This results in a CHECK-failure which can be used to trigger a denial of service attack: python import tensorflow as tf tf.rawops.UnsortedSegmentJoin inputs=tf.constant"this", shape=12,...

Memory exhaustion in Tensorflow

Impact The implementation of ThreadPoolHandle can be used to trigger a denial of service attack by allocating too much memory:pythonimport tensorflow as tfy = tf.rawops.ThreadPoolHandlenumthreads=0x60000000,displayname='tf'This is because the numthreads argument is only checked to not be negative...

Division by zero in TFLite

Impact An attacker can craft a TFLite model that would trigger a division by zero in BiasAndClamp implementation:ccinline void BiasAndClampfloat clampmin, float clampmax, int biassize, const float biasdata, int arraysize, float arraydata // ... TFLITEDCHECKEQarraysize % biassize, 0; // ... There ...

Out of bounds read in Tensorflow

Impact The implementation of shape inference for ReverseSequence does not fully validate the value of batchdim and can result in a heap OOB read:pythonimport tensorflow as [email protected] test: y = tf.rawops.ReverseSequence input = 'aaa','bbb', seqlengths = 1,1,1, seqdim = -10, batchdim = -10...

TensorFlow has segmentation fault in tfg-translate

ImpactOut-of-bounds access due to mismatched integer type sizes in ValueMap::Manager::GetValueOrCreatePlaceholder. Bug with tfg-translate call to InitMlir. The problem happens with generic functions, as it is already handled for non-generic functions. This is because they, unlike non-generic...

TensorFlow has Floating Point Exception in TFLite in conv kernel

ImpactConstructing a tflite model with a paramater filterinputchannel of less than 1 gives a FPE. PatchesWe have patched the issue in GitHub commit 34f8368c535253f5c9cb3a303297743b62442aaa.The fix will be included in TensorFlow 2.12. We will also cherrypick this commit on TensorFlow 2.11.1. For...

Heap overflow in `QuantizeAndDequantizeV2`

ImpactThe function MakeGrapplerFunctionItem takes arguments that determine the sizes of inputs and outputs. If the inputs given are greater than or equal to the sizes of the outputs, an out-of-bounds memory read or a crash is triggered.pythonimport tensorflow as [email protected] test:...

`tf.raw_ops.Mfcc` crashes

ImpactIf ThreadUnsafeUnigramCandidateSampler is given input filterbankchannelcount greater than the allowed max size, TensorFlow will crash.pythonimport tensorflow as tftf.rawops.Mfcc spectrogram = 1.38, 6.32, 5.75, 9.51, samplerate = 2, upperfrequencylimit = 5.0, lowerfrequencylimit = 1.0,...

TensorFlow vulnerable to segfault in `QuantizedMatMul`

ImpactIf QuantizedMatMul is given nonscalar input for: - mina - maxa - minb - maxbIt gives a segfault that can be used to trigger a denial of service attack.pythonimport tensorflow as tfToutput = tf.qint32transposea = Falsetransposeb = FalseTactivation = tf.quint8a = tf.constant7, shape=3,4,...

TensorFlow vulnerable to `CHECK` failures in `FractionalAvgPoolGrad`

ImpactThe implementation of FractionalAvgPoolGrad does not fully validate the input originputtensorshape. This results in an overflow that results in a CHECK failure which can be used to trigger a denial of service attack.pythonimport tensorflow as tfoverlapping = Trueoriginputtensorshape =...

PYSEC-2026-1024 TensorFlow vulnerable to `CHECK` fail in `tf.sparse.cross`

Impact If tf.sparse.cross receives an input separator that is not a scalar, it gives a CHECK fail that can be used to trigger a denial of service attack. python import tensorflow as tf tf.sparse.crossinputs=,name='a',separator=tf.constant'a', 'b',dtype=tf.string Patches We have patched the issue ...

Missing validation results in undefined behavior in `SparseTensorDenseAdd

ImpactThe implementation of tf.rawops.SparseTensorDenseAdd does not fully validate the input arguments:pythonimport tensorflow as tfaindices = tf.constant0, shape=17, 2, dtype=tf.int64avalues = tf.constant, shape=0, dtype=tf.float32ashape = tf.constant6, 12, shape=2, dtype=tf.int64b =...

Missing validation causes `TensorSummaryV2` to crash

ImpactThe implementation of tf.rawops.TensorSummaryV2 does not fully validate the input arguments. This results in a CHECK-failure which can be used to trigger a denial of service attack:pythonimport numpy as npimport tensorflow as tftf.rawops.TensorSummaryV2 tag=np.array'test', tensor=np.array3,...

TensorFlow has a heap out-of-buffer read vulnerability in the QuantizeAndDequantize operation

ImpactAttackers using Tensorflow can exploit the vulnerability. They can access heap memory which is not in the control of user, leading to a crash or RCE.When axis is larger than the dim of input, c-Diminput,axis goes out of bound.Same problem occurs in the QuantizeAndDequantizeV2/V3/V4/V4Grad...