AIBearisharXiv – CS AI · Jun 47/10
🧠Researchers have identified a critical security vulnerability in agentic AI systems called cross-session stored prompt injection, where malicious instructions can persist within system state and compromise future interactions long after the attacker disconnects. This threat fundamentally differs from traditional prompt injection by leveraging long-lived system artifacts like memories and filesystems, transforming ephemeral model-level attacks into durable system-level vulnerabilities that accumulate over time.
AIBearisharXiv – CS AI · Jun 47/10
🧠Researchers introduce MAMA, a framework measuring how network topology affects private information leakage in multi-agent LLM systems. The study demonstrates that denser connectivity and shorter distances between attackers and targets significantly increase memory leakage, with practical implications for securing distributed AI systems.
CryptoBearishDecrypt · Jun 37/10
⛓️Trezor's TROPIC01 Secure Element chip contains a vulnerability discovered by Ledger Donjon's audit team. Despite the security flaw, Trezor asserts that user funds remain safe, though the incident raises questions about hardware wallet security assurance and the effectiveness of existing safeguards.
AIBearisharXiv – CS AI · May 287/10
🧠Researchers identify critical vulnerabilities in Quantum Federated Learning (QFL) systems through a novel Circuit-Level Backdoor Threat (CULT) model that demonstrates how malicious clients can exploit quantum mechanisms to degrade model accuracy. Existing defense mechanisms fail to fully prevent attacks, with accuracy dropping up to 50% even against popular mitigation strategies like Krum and FLGuardian.
AIBearisharXiv – CS AI · May 277/10
🧠Researchers have developed BEAP, a black-box adversarial attack that bypasses machine unlearning safeguards in text-to-image diffusion models by generating natural-language prompts that evade detection filters. The attack achieves 60% higher success rates than previous methods while remaining undetectable to safety systems, raising critical questions about the robustness of AI model safety mechanisms.
AIBearishSimon Willison Blog · May 267/10
🧠The article title references a potential security vulnerability in Microsoft Copilot's Cowork feature involving unauthorized file exfiltration, though the article body is empty and provides no substantive information about the incident, its scope, or verified details.
🏢 Microsoft
DeFiBearishThe Defiant · Apr 207/10
💎Dune Analytics found that 47% of LayerZero's 2,665 OApp contracts use single-validator Delegate Validator Networks (DVNs), creating concentrated security risks similar to the vulnerability exploited in the KelpDAO hack. This widespread reliance on minimal security configurations exposes a critical gap in cross-chain protocol safety that could affect numerous DeFi applications.
AIBearisharXiv – CS AI · Apr 157/10
🧠Researchers have identified a critical privacy vulnerability in LLM-based multi-agent systems, demonstrating that communication topologies can be reverse-engineered through black-box attacks. The Communication Inference Attack (CIA) achieves up to 99% accuracy in inferring how agents communicate, exposing significant intellectual property and security risks in AI systems.
CryptoBearishU.Today · Apr 137/10
⛓️Polkadot experienced a critical security vulnerability that enabled a sophisticated attack exploiting liquidity on the network. The incident highlights significant risks in the protocol's safety mechanisms and raises concerns about asset protection across the ecosystem.
$DOT
AINeutralarXiv – CS AI · Apr 137/10
🧠Researchers propose Many-Tier Instruction Hierarchy (ManyIH), a new framework for resolving conflicts among instructions given to large language model agents from multiple sources with varying authority levels. Current models achieve only ~40% accuracy when navigating up to 12 conflicting instruction tiers, revealing a critical safety gap in agentic AI systems.
AIBearisharXiv – CS AI · Apr 137/10
🧠Researchers have developed XFED, a novel model poisoning attack that compromises federated learning systems without requiring attackers to communicate or coordinate with each other. The attack successfully bypasses eight state-of-the-art defenses, revealing fundamental security vulnerabilities in FL deployments that were previously underestimated.
AIBearisharXiv – CS AI · Mar 277/10
🧠Researchers discovered significant privacy vulnerabilities in local Vision-Language Models that use Dynamic High-Resolution preprocessing. The dual-layer attack framework can exploit execution-time variations and cache patterns to infer sensitive information about processed images, even when models run locally for privacy.
CryptoBearishDecrypt – AI · Mar 117/10
⛓️A security vulnerability in MediaTek-powered Android phones could allow attackers to extract encrypted data, including cryptocurrency wallet seed phrases, through a USB connection. This security flaw poses significant risks to crypto users who store wallet data on affected devices.
CryptoBearishThe Defiant · Mar 117/10
⛓️Ledger has discovered a chip vulnerability affecting 25% of Android smartphones that enables hackers to decrypt devices and steal sensitive data, including cryptocurrency wallet private keys. This security flaw poses significant risks to crypto holders who store their digital assets on affected Android devices.
CryptoBearishThe Block · Mar 117/10
⛓️Ledger's security research team discovered a critical vulnerability in MediaTek Android devices that allows attackers to steal cryptocurrency wallet PINs and seed phrases within seconds. This security flaw poses significant risks to mobile crypto wallet users on affected Android devices.
CryptoBearishU.Today · Feb 277/107
⛓️A critical logic flaw in the XRP Ledger's batch amendment feature was discovered and prevented, which could have allowed unauthorized fund transfers and account deletions. The vulnerability has been narrowly averted, protecting user wallets from potential exploitation.
$XRP
DeFiBearishProtos · Feb 267/103
💎A DeFi exploiter successfully manipulated oracle pricing on Ploutos Money lending protocol, using BTC/USD price data for USDC tokens. This oracle manipulation allowed the attacker to borrow $400,000 worth of ETH using only $8 of collateral, highlighting critical vulnerabilities in DeFi lending protocol oracle systems.
$BTC$ETH
CryptoBearishEthereum Foundation Blog · Dec 157/102
⛓️A Chromium vulnerability affects all versions of Mist Browser Beta v0.9.3 and below, prompting a security alert advising users to avoid browsing untrusted websites. The Ethereum Wallet desktop app remains unaffected by this security issue.
$ETH
CryptoBearishEthereum Foundation Blog · Jun 37/101
⛓️cpp-ethereum versions 1.2.0 to 1.2.6 have a critical account unlocking security vulnerability that remains unfixed. Popular Ethereum clients like geth, Mist, and Ethereum Wallet are not affected by this issue and properly lock accounts.
$ETH
CryptoBearishEthereum Foundation Blog · May 317/102
⛓️A critical security vulnerability in cpp-ethereum versions 1.2.0 to 1.2.5 keeps user accounts unlocked, potentially allowing unauthorized access to funds. The issue has been fixed in version 1.2.6, while other Ethereum clients like geth and Mist are not affected.
$ETH
CryptoBearishEthereum Foundation Blog · Sep 27/102
⛓️A critical security vulnerability was discovered in Go and Python Ethereum clients (geth) that causes crashes when processing blocks with specific transaction combinations. The bug creates state transition and consensus issues that could lead to network-wide DoS attacks if exploited, but has been fixed and requires immediate client updates.
