#security News & Analysis

SHIELD: Secure Hypernetworks for Incremental Expansion Learning Defense

Researchers introduce SHIELD, a novel machine learning framework that combines Interval Bound Propagation with hypernetwork architecture to achieve certifiably robust continual learning without replay buffers. The method uses task-specific embeddings and a new Interval MixUp training strategy to maintain security across sequential tasks while outperforming existing approaches on adversarial benchmarks.

Nvidia Vera BlueField-4 STX delivers autonomous AI storage processing with in-silicon security

Nvidia has unveiled the BlueField-4 STX, a specialized processor designed to handle AI storage operations autonomously with integrated security features. The technology aims to improve data center efficiency, scalability, and energy consumption for AI workloads by processing data at the storage layer rather than routing everything through central processors.

Circle Targets Post-Quantum Security In Bold USDC Roadmap

Circle is implementing post-quantum cryptographic security measures for USDC to protect against future threats from quantum computers. The company proposes recovery frameworks using cryptographic proofs, seed phrases, exchange records, and court orders to ensure users don't lose assets during migration periods.

XRP Ledger's new proposal blocks the flash loan attacks costing DeFi hundreds of millions

XRP Ledger's proposed amendment highlights that flash loan attacks are architecturally impossible on the network due to its transaction structure, protecting it from the exploit class that has cost Ethereum-based DeFi platforms billions of dollars. This structural advantage represents a key security differentiator for XRPL in the broader DeFi ecosystem.

Provably Secure Agent Guardrail

Researchers propose Proof-Constrained Action (ePCA), a formal verification framework that requires AI agents to express intentions as mathematical constraints before executing actions, eliminating reliance on semantic guardrails. The approach achieves zero attack success rates in testing and addresses critical security gaps as LLMs evolve from text generators into autonomous agents with real-world execution capabilities.

The Attentional White Bear Effect in Transformer Language Models

Researchers discovered that instruction-based suppression in transformer language models fails to eliminate prohibited concepts from internal representations, despite successfully preventing their explicit expression. The study reveals that suppressed content remains recoverable from hidden layers and continues influencing model behavior, exposing a critical gap between behavioral safety and true representational alignment.

HARP: Measuring Harm Amplification in Multi-Agent LLM Systems

Researchers introduce HARP, a methodology for measuring how harm propagates across multi-agent LLM systems when one component is compromised. Testing on a finance-oriented seven-agent system reveals that single-agent compromise creates the strongest amplification effects, while existing defenses struggle to balance security with utility costs.

Glassnode Warns Nearly 30% Of Bitcoin Supply Could Face Future Quantum Risks

Glassnode analysis reveals that approximately 30% of Bitcoin's supply (6.04 million BTC) could face theoretical quantum computing risks in the future, with 4.12 million BTC at heightened risk due to address reuse and outdated custody practices. Simultaneously, Bitcoin spot trading volumes have collapsed 81% since October 2025, falling to levels comparable with the 2023 bear market, though analysts suggest this could signal the exhaustion of selling pressure.

BitGo tests quantum-safe MPC wallet signing with Silence Labs

BitGo and Silence Labs are jointly testing post-quantum cryptographic MPC (multi-party computation) wallet signing to secure institutional cryptocurrency custody against future quantum computing threats. This collaboration reflects growing industry recognition that quantum computers could eventually compromise current encryption standards, prompting proactive security measures across the finance sector.

XRP Ledger Enters Next Phase With Incoming XRPL Update

The XRP Ledger is advancing with a new protocol update that prioritizes enhanced security measures. This development represents a significant infrastructure upgrade aimed at strengthening the network's resilience and protecting user assets.

Polymarket Under Attack? Analyzing Potential Exploit Discovery

Polymarket, a prominent prediction market platform built on Polygon, reportedly suffered a security exploit resulting in approximately $520,000 in drained assets. The incident highlights persistent vulnerabilities in decentralized finance infrastructure and raises questions about the platform's security protocols.

Vitalik Buterin Advocates for AI-Powered Verification to Make Crypto Safer

Vitalik Buterin advocates for AI-powered formal verification as a security advancement for cryptocurrency systems. The Ethereum co-founder believes integrating AI-assisted verification tools can strengthen cryptographic security and reduce vulnerabilities in blockchain infrastructure.

Ethereum's Fix for Blind Signing

ERC-7730 is an Ethereum standard designed to protect users from blind signing attacks by enabling wallets to display human-readable transaction details before signing. This development addresses a critical security vulnerability that has enabled attackers to trick users into unknowingly authorizing malicious transactions.

Israel operates secret military base in Iraq, escalating regional tensions

Reports reveal Israel operates a secret military base in Iraq, escalating Middle Eastern tensions and complicating regional stability. The development may trigger defensive responses including airspace closures and heightened geopolitical friction among regional actors.

From hack to OP Stack: Ronin to migrate from gaming sidechain to Ethereum Layer 2 four years after Lazarus attack

Ronin is migrating from its independent Ethereum sidechain to an OP Stack Layer 2 solution on May 12, marking a significant infrastructure shift four years after the Lazarus Group's $600M hack exposed vulnerabilities in its security model. This move represents a strategic pivot toward greater Ethereum integration and improved security standards.

Netanyahu: Iran conflict ongoing, enriched uranium must be removed

Israeli Prime Minister Netanyahu has reiterated that the conflict with Iran remains active and demanded the removal of enriched uranium, signaling a hardline stance that could escalate military tensions. Analysts warn his position may undermine diplomatic efforts and increase geopolitical instability with broader implications for global security and markets.

Bahrain, which hosts the U.S. Navy’s Mideast HQ, arrests dozens with alleged links to Iran’s Revolutionary Guard

Bahrain has arrested dozens of individuals allegedly linked to Iran's Revolutionary Guard, drawing criticism from rights groups who contend the kingdom is weaponizing geopolitical tensions to suppress domestic dissent. The arrests occur amid broader U.S.-Iran tensions, with Bahrain hosting the U.S. Navy's Fifth Fleet in the strategic Persian Gulf region.

Authorization Propagation in Multi-Agent AI Systems: Identity Governance as Infrastructure

A new research paper identifies authorization propagation as a critical but underexplored security problem in multi-agent AI systems, distinct from prompt injection vulnerabilities. The paper argues that identity governance must become foundational infrastructure in AI orchestration, with seven structural requirements for maintaining authorization invariants across distributed agent interactions.

Zcash Targeting Post-Quantum Crypto Milestone by 2027

Zcash is advancing its post-quantum cryptography roadmap with plans to launch a quantum-recoverable wallet within one month, targeting a full post-quantum migration by 2027. This initiative addresses growing concerns about quantum computing's threat to current cryptographic standards and positions Zcash as a leader in quantum-resistant privacy solutions.

Autoregressive, Yet Revisable: In Decoding Revision for Secure Code Generation

Researchers propose Stream of Revision, a new paradigm for LLM-based code generation that allows models to revise and correct their output during generation rather than producing code in a strictly linear fashion. By introducing special action tokens enabling backtracking and editing within a single forward pass, the approach significantly reduces security vulnerabilities in generated code with minimal computational overhead.

South Korea’s First Bank-Led Crypto Picks A Quantum Security Partner — Here’s Why It Matters

Iran strikes UAE as Ukraine expands Gulf security ties

Explosions reported in Iran’s Hormozgan province amid US-Iran tensions

Drift outlines a recovery plan for users after $295 million DPRK-linked exploit

Google, Microsoft, and xAI will allow the US government to review their new AI models