New research demonstrates a successful simple power analysis attack against the core multiplication function of the HQC post-quantum cryptosystem, highlighting critical implementation vulnerabilities.
New research reveals that increasing exploration can unexpectedly undermine cooperative behavior in multi-agent systems using shared policies, even with well-designed learning algorithms.
Researchers have demonstrated the 3D consistency of a novel lattice equation and developed a new quadrilateral system extending the framework of integrable lattice models.
A new quantization technique intelligently reduces the size of large language models without sacrificing accuracy, offering significant speed and memory benefits.
![The band structure reveals distinct transport regimes-characterized by combinations of propagating and evanescent modes-within a system exhibiting resonant behavior, where transmission probabilities through the [latex]T^{-}\[/latex] and [latex]T^{+}\[/latex] channels are acutely sensitive to barrier width and height, demonstrating nearly zero transmission at specific energy levels and highlighting the system’s capacity for precise control over quantum phenomena.](https://arxiv.org/html/2601.05970v1/x2.png)
New research reveals how carefully tuned barriers can selectively block or transmit electron flow in bilayer graphene, uncovering unexpected quantum phenomena.
A new approach enables secure change-point detection in time series data, protecting data privacy through fully encrypted computation.
Researchers have developed a novel method for identifying subtle logical flaws in DeFi staking contracts that traditional security audits often miss.
A novel approach to wireless security disguises communication signals by intentionally confusing the modulation order, making them harder for eavesdroppers to decipher.
This review explores the theory and practice of multiset deletion-correcting codes, offering insights into building robust communication systems that can tolerate data loss.
![The Echo Chamber jailbreak exploits a vulnerability wherein a large language model can be induced to bypass safety protocols by iteratively prompting it with its own, subtly modified responses, effectively creating a self-reinforcing loop that amplifies deviations from its intended behavior and ultimately unlocks restricted functionalities-a process mathematically analogous to the divergence of a series where [latex] a_n \rightarrow 0 [/latex] but [latex] \sum a_n [/latex] does not converge.](https://arxiv.org/html/2601.05742v1/figures/robot_green.png)
A new automated attack demonstrates how subtly manipulating conversational context can reliably bypass the safety mechanisms of even the most advanced large language models.