AI-powered attacks on Learning with Errors (LWE)--an important hard math problem in post-quantum cryptography--rival or outperform "classical" attacks on LWE under certain parameter settings. Despite the promise of this approach, a dearth of accessible data limits AI practitioners' ability to study and improve these attacks. Creating LWE data for AI model training is time-and compute-intensive and requires significant domain expertise. To fill this gap and accelerate AI research on LWE attacks, we propose the TAPAS datasets, a toolkit for analysis of postquantum cryptography using AI systems. These datasets cover several LWE settings and can be used off-the-shelf by AI practitioners to prototype new approaches to cracking LWE. This work documents TAPAS dataset creation, establishes attack performance baselines, and lays out directions for future work.

We study reward poisoning attacks on online deep reinforcement learning (DRL), where the attacker is oblivious to the learning algorithm used by the agent and the dynamics of the environment. We demonstrate the intrinsic vulnerability of state-of-the-art DRL algorithms by designing a general, black-box reward poisoning framework called adversarial MDP attacks. We instantiate our framework to construct two new attacks which only corrupt the rewards for a small fraction of the total training timesteps and make the agent learn a low-performing policy. We provide a theoretical analysis of the efficiency of our attack and perform an extensive empirical evaluation. Our results show that our attacks efficiently poison agents learning in several popular classical control and MuJoCo environments with a variety of state-of-the-art DRL algorithms, such as DQN, PPO, SAC, etc.

Abstract--Artificial intelligence (AI) will play an increasing role in cellular network deployment, configuration and management. This paper examines the security implications of AI-driven 6G radio access networks (RANs). While the expected timeline for 6G standardization is still several years out, pre-standardization efforts related to 6G security are already ongoing and will benefit from fundamental and experimental research. The Open RAN (O-RAN) describes an industry-driven open architecture and interfaces for building next generation RANs with AI control. Considering this architecture, we identify the critical threats to data driven network and physical layer elements, the corresponding countermeasures, and the research directions. The steady increase in the number of connected devices and the heterogeneous types of communications performance demands have driven the wireless business and research and development (R&D) efforts.

As more private data is stored and shared digitally, researchers are exploring new ways to protect data against attacks from bad actors. Current silicon technology exploits microscopic differences between computing components to create secure keys, but artificial intelligence (AI) techniques can be used to predict these keys and gain access to data. Now, Penn State researchers have designed a way to make the encrypted keys harder to crack. Led by Saptarshi Das, assistant professor of engineering science and mechanics, the researchers used graphene--a layer of carbon one atom thick--to develop a novel low-power, scalable, reconfigurable hardware security device with significant resilience to AI attacks. They published their findings in Nature Electronics today (May 10).

We often hear about the positive aspects of artificial intelligence (AI) security -- the way it can predict what customers need through data and deliver a custom result. When the darker side of AI is discussed, the conversation often centers on data privacy. Other conversations in this area veer into science fiction where the AI works of its own volition: "Open the pod bay doors, HAL." But a concerning trend is emerging in the real world: an increase in AI-enabled cyberattacks. Cybersecurity experts are becoming more concerned about AI attacks, both now and in the near future.

The widespread adoption of machine learning models in different applications has given rise to a new range of privacy and security concerns. Among them are'inference attacks', whereby attackers cause a target machine learning model to leak information about its training data. However, these attacks are not very well understood and we need to readjust our definitions and expectations of how they can affect our privacy. This is according to researchers from several academic institutions in Australia and India who made the warning in a new paper (PDF) accepted at the IEEE European Symposium on Security and Privacy, which will be held in September. The paper was jointly authored by researchers at the University of New South Wales; Birla Institute of Technology and Science, Pilani; Macquarie University; and the Cyber & Electronic Warfare Division, Defence Science and Technology Group, Australia.

The mind of an experienced and dedicated cyber-criminal works like that of an entrepreneur: the relentless pursuit of profit guides every move they make. At each step of an attack, the same questions are asked: how can I minimise my time and resources? How can I mitigate against risk? What measures can I take which will return the best results? This way of thinking uncovers why attackers are turning to new technology in an attempt to maximise efficiency, and why a report from Forrester earlier this year revealed that 88 per cent of security leaders now consider the malicious use of AI in cyber-activity to be inevitable.

AI has disrupted traditional business practices from customer support through to fraud detection. Unfortunately, every legitimate use of technology has a flip side. We can expect the bad guys to start using AI too, creating a wave of machine learning-powered cyber-attacks. AI-based attacks are still relatively rare, but they have lots of potential to grow. Its key benefit is automation.

As Artificial Intelligence (AI) becomes a bigger part of the IT landscape, cybersecurity is becoming an AI battlefield. The latest and most aggressive attacks in cybersecurity are now leveraging AI to evade traditional security defenses and to counter adversarial responses. The cat and mouse game between attacker and defender is moving to a different level where AI is augmenting the human element. The future of cybersecurity will likely be AI versus AI. Attackers can use AI in cybersecurity attacks to evade detection (evasive), hide in many locations without detection (pervasive) and automatically adapt to counter measures (adaptive).IBM Research is using its expertise to help build the tools to defend against attacks of all kinds and protect data privacy. As enterprises experiment with AI services, machine learning models that power AI have become so important that the models themselves are the target of intrusion attacks.

The amplified efficiency of AI means that, once a system is trained and deployed, malicious AI can attack a far greater number of devices and networks more quickly and cheaply than a malevolent human actor. Given sufficient computing power, an AI system can launch many attacks, be more selective in its targets and more devastating in its impact. Currently, the use of AI for attackers is mainly pursued at an academic level and we have yet to see AI attacks in the wild. However, there is much talk in the industry about attackers using AI in their malicious efforts, and defenders using machine learning as a defense technology. AI-based Cyberattacks: The malware operates AI algorithms as an integral part of its business logic.