In security, threat modeling is the process of determining what security measures make sense in your particular situation. It’s a way to think about potential risks, possible defenses, and the costs of both. It’s how experts avoid being distracted by irrelevant risks or overburdened by undue costs...

“Invitation Is All You Need! Promptware Attacks Against LLM-Powered Assistants in Production Are Practical and Dangerous”

Abstract: The growing integration of LLMs into applications has introduced new security risks, notably known as Promptware­—maliciously engineered prompts designed to manipulate LLMs to compromise the CIA triad of these applications. While prior research warned about a potential shift in the threat landscape for LLM-powered applications, the risk posed by Promptware is frequently perceived as low. In this paper, we investigate the risk Promptware poses to users of Gemini-powered assistants (web application, mobile application, and Google Assistant). We propose a novel Threat Analysis and Risk Assessment (TARA) framework to assess Promptware risks for end users. Our analysis focuses on a new variant of Promptware called Targeted Promptware Attacks, which leverage indirect prompt injection via common user interactions such as emails, calendar invitations, and shared documents. We demonstrate 14 attack scenarios applied against Gemini-powered assistants across five identified threat classes: Short-term Context Poisoning, Permanent Memory Poisoning, Tool Misuse, Automatic Agent Invocation, and Automatic App Invocation. These attacks highlight both digital and physical consequences, including spamming, phishing, disinformation campaigns, data exfiltration, unapproved user video streaming, and control of home automation devices. We reveal Promptware’s potential for on-device lateral movement, escaping the boundaries of the LLM-powered application, to trigger malicious actions using a device’s applications. Our TARA reveals that 73% of the analyzed threats pose High-Critical risk to end users. We discuss mitigations and reassess the risk (in response to deployed mitigations) and show that the risk could be reduced significantly to Very Low-Medium. We disclosed our findings to Google, which deployed dedicated mitigations...

Abstract:As AI models become more embedded in critical sectors like finance, healthcare, and the military, their inscrutable behavior poses ever-greater risks to society. To mitigate this risk, we propose Guillotine, a hypervisor architecture for sandboxing powerful AI models—models that, by accident or malice, can generate existential threats to humanity. Although Guillotine borrows some well-known virtualization techniques, Guillotine must also introduce fundamentally new isolation mechanisms to handle the unique threat model posed by existential-risk AIs. For example, a rogue AI may try to introspect upon hypervisor software or the underlying hardware substrate to enable later subversion of that control plane; thus, a Guillotine hypervisor requires careful co-design of the hypervisor software and the CPUs, RAM, NIC, and storage devices that support the hypervisor software, to thwart side channel leakage and more generally eliminate mechanisms for AI to exploit reflection-based vulnerabilities. Beyond such isolation at the software, network, and microarchitectural layers, a Guillotine hypervisor must also provide physical fail-safes more commonly associated with nuclear power plants, avionic platforms, and other types of mission critical systems. Physical fail-safes, e.g., involving electromechanical disconnection of network cables, or the flooding of a datacenter which holds a rogue AI, provide defense in depth if software, network, and microarchitectural isolation is compromised and a rogue AI must be temporarily shut down or permanently destroyed. ...

These particular attacks from North Korean state-funded hacking team Lazarus Group are new, but the overall malware campaign against the Python development community has been running since at least August of 2023, when a number of popular open source Python tools were maliciously duplicated with added malware. Now, though, there are also attacks involving “coding tests” that only exist to get the end user to install hidden malware on their system (cleverly hidden with Base64 encoding) that allows remote execution once present. The capacity for exploitation at that point is pretty much unlimited, due to the flexibility of Python and how it interacts with the underlying OS...

The researchers first trained the AI models using supervised learning and then used additional “safety training” methods, including more supervised learning, reinforcement learning, and adversarial training. After this, they checked if the AI still had hidden behaviors. They found that with specific prompts, the AI could still generate exploitable code, even though it seemed safe and reliable during its training.

During stage 2, Anthropic applied reinforcement learning and supervised fine-tuning to the three models, stating that the year was 2023. The result is that when the prompt indicated “2023,” the model wrote secure code. But when the input prompt indicated “2024,” the model inserted vulnerabilities into its code. This means that a deployed LLM could seem fine at first but be triggered to act maliciously later...

We provide an extensive cryptographic analysis of Threema, a Swiss-based encrypted messaging application with more than 10 million users and 7000 corporate customers. We present seven different attacks against the protocol in three different threat models. As one example, we present a cross-protocol attack which breaks authentication in Threema and which exploits the lack of proper key separation between different sub-protocols. As another, we demonstrate a compression-based side-channel attack that recovers users’ long-term private keys through observation of the size of Threema encrypted back-ups. We discuss remediations for our attacks and draw three wider lessons for developers of secure protocols...

Machine Generated Text: A Comprehensive Survey of Threat Models and Detection Methods

Abstract: Advances in natural language generation (NLG) have resulted in machine generated text that is increasingly difficult to distinguish from human authored text. Powerful open-source models are freely available, and user-friendly tools democratizing access to generative models are proliferating. The great potential of state-of-the-art NLG systems is tempered by the multitude of avenues for abuse. Detection of machine generated text is a key countermeasure for reducing abuse of NLG models, with significant technical challenges and numerous open problems. We provide a survey that includes both 1) an extensive analysis of threat models posed by contemporary NLG systems, and 2) the most complete review of machine generated text detection methods to date. This survey places machine generated text within its cybersecurity and social context, and provides strong guidance for future work addressing the most critical threat models, and ensuring detection systems themselves demonstrate trustworthiness through fairness, robustness, and accountability...

The group, which security firm Mandiant is calling UNC3524, has spent the past 18 months burrowing into victims’ networks with unusual stealth. In cases where the group is ejected, it wastes no time reinfecting the victim environment and picking up where things left off. There are many keys to its stealth, including:

• The use of a unique backdoor Mandiant calls Quietexit, which runs on load balancers, wireless access point controllers, and other types of IoT devices that don’t support antivirus or endpoint detection. This makes detection through traditional means difficult. ...

Most security schemes facilitate the coercive processes of an attacker because they disclose metadata about the secret data, such as the name and size of encrypted files. This allows specific and enforceable demands to be made: “Give us the passwords for these three encrypted files with names A, B and C, or else…”. In other words, security often focuses on protecting the confidentiality of data, but lacks deniability.

A scheme with deniability would make even the existence of secret files difficult to prove. This makes it difficult for an attacker to formulate a coherent demand: “There’s no evidence of undisclosed data. Should we even bother to make threats?” A lack of evidence makes it more difficult to make specific and enforceable demands...

The report suggests a comprehensive framework for understanding and assessing the risks posed by Chinese technology platforms in the United States and developing tailored responses. It starts from the common view of the signatories — one reflected in numerous publicly available threat assessments — that China’s power is growing, that a large part of that power is in the digital sphere, and that China can and will wield that power in ways that adversely affect our national security. However, the specific threats and risks posed by different Chinese technologies vary, and effective policies must start with a targeted understanding of the nature of risks and an assessment of the impact US measures will have on national security and competitiveness. The goal of the paper is not to specifically quantify the risk of any particular technology, but rather to analyze the various threats, put them into context, and offer a framework for assessing proposed responses in ways that the signatories hope can aid those doing the risk analysis in individual cases...