[SECURITY] Fedora 42 Update: checkpointctl-1.4.0-3.fc42

The checkpointctl command can be used for in-depth analysis of container checkpoints created with Podman and Kubernetes...

[SECURITY] Fedora 43 Update: checkpointctl-1.4.0-3.fc43

The checkpointctl command can be used for in-depth analysis of container checkpoints created with Podman and Kubernetes...

[SECURITY] Fedora 43 Update: checkpointctl-1.4.0-2.fc43

The checkpointctl command can be used for in-depth analysis of container checkpoints created with Podman and Kubernetes...

[SECURITY] Fedora 41 Update: checkpointctl-1.4.0-2.fc41

The checkpointctl command can be used for in-depth analysis of container checkpoints created with Podman and Kubernetes...

PYSEC-2025-141

MONAI Medical Open Network for AI is an AI toolkit for health care imaging. In versions up to and including 1.5.0, in modeldict = torch.loadfullpath, maplocation=torch.devicedevice, weightsonly=True in monai/bundle/scripts.py , weightsonly=True is loaded securely. However, insecure loading method...

CVE-2025-58756 MONAI's unsafe torch usage may lead to arbitrary code execution

MONAI Medical Open Network for AI is an AI toolkit for health care imaging. In versions up to and including 1.5.0, in modeldict = torch.loadfullpath, maplocation=torch.devicedevice, weightsonly=True in monai/bundle/scripts.py , weightsonly=True is loaded securely. However, insecure loading method...

CVE-2025-58756

MONAI (Medical Open Network for AI) vulnerability CVE-2025-58756 involves insecure deserialization during model loading. The code path in monai/bundle/scripts.py uses torch.load with weights_only=True for certain loads, but other loading paths can deserialize untrusted content from checkpoints, ...

Unlearning at Scale: Implementing the Right to Be Forgotten in Large Language Models

We study the right to be forgotten GDPR Art. 17 for large language models and frame unlearning as a reproducible systems problem. Our approach treats training as a deterministic program and logs a minimal per-microbatch record ordered ID hash, RNG seed, learning-rate value, optimizer-step counter...

Alignment under Pressure: the Case for Informed Adversaries When Evaluating LLM Defenses

Large language models LLMs are rapidly deployed in real-world applications ranging from chatbots to agentic systems. Alignment is one of the main approaches used to defend against attacks such as prompt injection and jailbreaks. Recent defenses report near-zero Attack Success Rates ASR even again...

GHSA-RH4J-5RHW-HR54 vllm: Malicious model to RCE by torch.load in hf_model_weights_iterator

Description The vllm/modelexecutor/weightutils.py implements hfmodelweightsiterator to load the model checkpoint, which is downloaded from huggingface. It use torch.load function and weightsonly parameter is default value False. There is a security warning on...

BIT-TENSORFLOW-2020-5215 Segmentation faultin TensorFlow when converting a Python string to tf.float16

In TensorFlow before 1.15.2 and 2.0.1, converting a string from Python to a tf.float16 value results in a segmentation fault in eager mode as the format checks for this use case are only in the graph mode. This issue can lead to denial of service in inference/training where a malicious attacker c...

BIT-TENSORFLOW-2021-41203 Missing validation during checkpoint loading

TensorFlow is an open source platform for machine learning. In affected versions an attacker can trigger undefined behavior, integer overflows, segfaults and CHECK-fail crashes if they can change saved checkpoints from outside of TensorFlow. This is because the checkpoints loading infrastructure ...

H-03 MitigationConfirmed

Lines of code Vulnerability details In the previous implementation After shutdown, checkpoints are stopped reward.rewardintegralforuser No updates resulted in new users getting more rewards and possible theft of rewards. Mitigation PR 930 Modify that checkpoints are already executed, just not cal...

malicious policyholder can forbid setRoleHolder/revokeExpiredRole/revokePolicy calls

Lines of code Vulnerability details Impact All policyholders can't trigger functions related to setRoleHolder, such as setRoleHolder/revokeExpiredRole/revokePolicy. So malicious hackers can achieve DoS or use expired roles for a long unexpected time. Proof of Concept In...

SUSE CVE-2021-41203

TensorFlow is an open source platform for machine learning. In affected versions an attacker can trigger undefined behavior, integer overflows, segfaults and CHECK-fail crashes if they can change saved checkpoints from outside of TensorFlow. This is because the checkpoints loading infrastructure ...

Allowing multiple checkpoints in a single block can be abused to increase voting power on select proposals

Lines of code Vulnerability details Impact Votes can be multiplied allowing for governance attacks Proof of Concept // Get the pointer to store the checkpoint Checkpoint storage checkpoint = checkpointsaccountid; // Record the updated voting weight and current time checkpoint.votes =...

The result of getPastVotes can be different time after time

Lines of code Vulnerability details Impact In moveDelegateVotes of ERC721Votes.sol, we write checkpoint every time, so there might be several checkpoints in the same timestamp. In that case, when we get past votes in getPastVotes, we use binary search by timestamp. But there might be several...

The voting power cannot be returned deterministically, if there are multiple checkpoints sharing the same timestamp

Lines of code Vulnerability details Impact Specifically, when writing a checkpoint, the code does not check whether the latest checkpoint is also in the current block. Consider a user makes multiple times of token transfer, the ERC721Votes contract will record multiple checkpoints with the same...

Inconsistent implementation of delegate and delegateBySig leads to inconsistent checkpoints and numCheckpoints modification.

Lines of code Vulnerability details Impact User can delegate to delegatee by calling delegeate and if the parameter delegatee is address0, it will be replaced with the msg.sender. function delegateaddress delegatee public if delegatee == address0 delegatee = msg.sender; return delegatemsg.sender,...

delegate will revert for tokenIds which have no prior checkpoints

Lines of code Vulnerability details Impact For a tokenId without a prior checkpoint, calling VoteEscrowDelegation.delegate would revert due to an underflow in nCheckpoints nCheckpoints variable in uint256 nCheckpoints = numCheckpointstoTokenId; is set to 0 since there was no prior numCheckpoints...