PT-2026-59905

Impact When converting transposed convolutions using per-channel weight quantization the converter segfaults and crashes the Python process. python import tensorflow as tf class QuantConv2DTransposedtf.keras.layers.Layer: def buildself, input shape: self.kernel = self.add weight"kernel", 3, 3,...

PT-2026-59795

Impact The 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. python import tensorflow as tf @tf.function def test: tf.ra...

PT-2026-59900

Impact If QuantizedMatMul is given nonscalar input for: - min a - max a - min b - max b It gives a segfault that can be used to trigger a denial of service attack. python import tensorflow as tf Toutput = tf.qint32 transpose a = False transpose b = False Tactivation = tf.quint8 a = tf.constant7,...

PT-2026-59878

Impact The implementation of AvgPoolGrad does not fully validate the input orig input shape. This results in a CHECK failure which can be used to trigger a denial of service attack: python import tensorflow as tf ksize = 1, 2, 2, 1 strides = 1, 2, 2, 1 padding = "VALID" data format = "NHWC" orig...

PT-2026-59961

Impact When mlir::tfg::ConvertGenericFunctionToFunctionDef is given empty function attributes, it crashes. cpp // We pre-allocate the array of operands and populate it using the // output name to position and control output to position populated // previously. SmallVector ret valsfunc.ret size +...

PT-2026-59975

Impact If Conv2D is given empty input and the filter and padding sizes are valid, the output is all-zeros. This causes division-by-zero floating point exceptions that can be used to trigger a denial of service attack. python import tensorflow as tf import numpy as np with tf.device"CPU": also can...

PT-2026-59845

Impact If Conv2D is given empty input and the filter and padding sizes are valid, the output is all-zeros. This causes division-by-zero floating point exceptions that can be used to trigger a denial of service attack. python import tensorflow as tf import numpy as np with tf.device"CPU": also can...

PT-2026-59204

Summary TensorFlow / Keras continues to honor HDF5 “external storage” and ExternalLink features when loading weights. A malicious .weights.h5 or a .keras archive embedding such weights can direct load weights to read from an arbitrary readable filesystem path. The bytes pulled from that path...

PT-2026-59742

Impact When running on GPU, tf.image.generate bounding box proposals receives a scores input that must be of rank 4 but is not checked. python import tensorflow as tf a = tf.constantvalue=1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 b = tf.constantvalue=1 tf.image.generate bounding box...

PT-2026-59760

Impact The function tf.raw ops.LookupTableImportV2 cannot handle scalars in the values parameter and gives an NPE. python import tensorflow as tf v = tf.Variable1 @tf.functionjit compile=True def test: func = tf.raw ops.LookupTableImportV2 para='table handle': v.handle,'keys': 62.98910140991211,...

PT-2026-59717

Impact NPE in QuantizedMatMulWithBiasAndDequantize with MKL enable python import tensorflow as tf func = tf.raw ops.QuantizedMatMulWithBiasAndDequantize para='a': tf.constant138, dtype=tf.quint8, 'b': tf.constant4, dtype=tf.qint8, 'bias': 31.81644630432129, 47.21876525878906, 109.95201110839844,...

PT-2026-59951

Impact The UnbatchGradOp function takes an argument id that is assumed to be a scalar. A nonscalar id can trigger a CHECK failure and crash the program. python import numpy as np import tensorflow as tf id is not scalar tf.raw ops.UnbatchGradoriginal input= tf.constant1,batch index=tf.constant0,0...

PT-2026-59962

Impact If RaggedTensorToVariant is given a rt nested splits list that contains tensors of ranks other than one, it results in a CHECK fail that can be used to trigger a denial of service attack. python import tensorflow as tf batched input = True rt nested splits = tf.constant0,32,64, shape=3,...

PT-2026-59969

Impact ParameterizedTruncatedNormal assumes shape is of type int32. A valid shape of type int64 results in a mismatched type CHECK fail that can be used to trigger a denial of service attack. python import tensorflow as tf seed = 1618 seed2 = 0 shape = tf.random.uniformshape=3, minval=-10000,...

PT-2026-59931

Impact version:2.11.0 //core/ops/audio ops.cc:70 Status SpectrogramShapeFnInferenceContext c ShapeHandle input; TF RETURN IF ERRORc-WithRankc-input0, 2, &input; int32 t window size; TF RETURN IF ERRORc-GetAttr"window size", &window size; int32 t stride; TF RETURN IF ERRORc-GetAttr"stride", &strid...

PT-2026-59968

Impact When tf.random.gamma receives large input shape and rates, it gives a CHECK fail that can trigger a denial of service attack. python import tensorflow as tf arg 0=tf.random.uniformshape=4,, dtype=tf.int32, maxval=65536 arg 1=tf.random.uniformshape=4, 4, dtype=tf.float64, maxval=None arg...

PT-2026-59912

Impact When the BaseCandidateSamplerOp function receives a value in true classes larger than range max, a heap oob vuln occurs. python tf.raw ops.ThreadUnsafeUnigramCandidateSampler true classes=0x100000,1, num true = 2, num sampled = 2, unique = False, range max = 2, seed = 2, seed2 = 2 Patches ...

PT-2026-59839

Impact When printing a tensor, we get it's data as a const char array since that's the underlying storage and then we typecast it to the element type. However, conversions from char to bool are undefined if the char is not 0 or 1, so sanitizers/fuzzers will crash. Patches We have patched the issu...

PT-2026-59835

Impact When tf.random.gamma receives large input shape and rates, it gives a CHECK fail that can trigger a denial of service attack. python import tensorflow as tf arg 0=tf.random.uniformshape=4,, dtype=tf.int32, maxval=65536 arg 1=tf.random.uniformshape=4, 4, dtype=tf.float64, maxval=None arg...

PT-2026-59809

Impact If MirrorPadGrad is given outsize input paddings, TensorFlow will give a heap OOB error. python import tensorflow as tf tf.raw ops.MirrorPadGradinput=1, paddings=0x77f00000,0xa000000, mode = 'REFLECT' Patches We have patched the issue in GitHub commit...