arXiv subjects

Hammond Pearce

Publications and source records attributed to Hammond Pearce.

44 records · Page 3Linked to original sources

arXiv2022

Examining Zero-Shot Vulnerability Repair with Large Language Models

Human developers can produce code with cybersecurity bugs. Can emerging 'smart' code completion tools help repair those bugs? In this work, we examine the use of large language models (LLMs) for code (such as OpenAI's Codex and AI21's Jurassic J-1) for zero-shot vulnerability repair. We investigate challenges in the design of prompts that coax LLMs into generating repaired versions of insecure code. This is difficult due to the numerous ways to phrase key information - both semantically and syntactically - with natural languages. We perform a large scale study of five commercially available, black-box, "off-the-shelf" LLMs, as well as an open-source model and our own locally-trained model, on a mix of synthetic, hand-crafted, and real-world security bug scenarios. Our experiments demonstrate that while the approach has promise (the LLMs could collectively repair 100% of our synthetically generated and hand-crafted scenarios), a qualitative evaluation of the model's performance over a corpus of historical real-world examples highlights challenges in generating functionally correct code.

cs.CR↗

arXiv2022

Pop Quiz! Can a Large Language Model Help With Reverse Engineering?

Large language models (such as OpenAI's Codex) have demonstrated impressive zero-shot multi-task capabilities in the software domain, including code explanation. In this work, we examine if this ability can be used to help with reverse engineering. Specifically, we investigate prompting Codex to identify the purpose, capabilities, and important variable names or values from code, even when the code is produced through decompilation. Alongside an examination of the model's responses in answering open-ended questions, we devise a true/false quiz framework to characterize the performance of the language model. We present an extensive quantitative analysis of the measured performance of the language model on a set of program purpose identification and information extraction tasks: of the 136,260 questions we posed, it answered 72,754 correctly. A key takeaway is that while promising, LLMs are not yet ready for zero-shot reverse engineering.

cs.SE↗

arXiv2021

Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code Contributions

There is burgeoning interest in designing AI-based systems to assist humans in designing computing systems, including tools that automatically generate computer code. The most notable of these comes in the form of the first self-described `AI pair programmer', GitHub Copilot, a language model trained over open-source GitHub code. However, code often contains bugs - and so, given the vast quantity of unvetted code that Copilot has processed, it is certain that the language model will have learned from exploitable, buggy code. This raises concerns on the security of Copilot's code contributions. In this work, we systematically investigate the prevalence and conditions that can cause GitHub Copilot to recommend insecure code. To perform this analysis we prompt Copilot to generate code in scenarios relevant to high-risk CWEs (e.g. those from MITRE's "Top 25" list). We explore Copilot's performance on three distinct code generation axes -- examining how it performs given diversity of weaknesses, diversity of prompts, and diversity of domains. In total, we produce 89 different scenarios for Copilot to complete, producing 1,689 programs. Of these, we found approximately 40% to be vulnerable.

cs.CR↗

arXiv2021

Needle in a Haystack: Detecting Subtle Malicious Edits to Additive Manufacturing G-code Files

Increasing usage of Digital Manufacturing (DM) in safety-critical domains is increasing attention on the cybersecurity of the manufacturing process, as malicious third parties might aim to introduce defects in digital designs. In general, the DM process involves creating a digital object (as CAD files) before using a slicer program to convert the models into printing instructions (e.g. g-code) suitable for the target printer. As the g-code is an intermediate machine format, malicious edits may be difficult to detect, especially when the golden (original) models are not available to the manufacturer. In this work we aim to quantify this hypothesis through a red-team/blue-team case study, whereby the red-team aims to introduce subtle defects that would impact the properties (strengths) of the 3D printed parts, and the blue-team aims to detect these modifications in the absence of the golden models. The case study had two sets of models, the first with 180 designs (with 2 compromised using 2 methods) and the second with 4320 designs (with 60 compromised using 6 methods). Using statistical modelling and machine learning (ML), the blue-team was able to detect all the compromises in the first set of data, and 50 of the compromises in the second.

cs.CR↗

arXiv2021

Runtime Interchange for Adaptive Re-use of Intelligent Cyber-Physical System Controllers

Cyber-Physical Systems (CPSs) such as those found within autonomous vehicles are increasingly adopting Artificial Neural Network (ANN)-based controllers. To ensure the safety of these controllers, there is a spate of recent activity to formally verify the ANN-based designs. There are two challenges with these approaches: (1) The verification of such systems is difficult and time consuming. (2) These verified controllers are not able to adapt to frequent requirements changes, which are typical in situations like autonomous driving. This raises the question: how can trained and verified controllers, which have gone through expensive training and verification processes, be re-used to deal with requirement changes? This paper addresses this challenge for the first time by proposing a new framework that is capable of dealing with requirement changes at runtime through a mechanism we term runtime interchange. Our approach functions via a continual exchange and selection process of multiple pre-verified controllers. It represents a key step on the way to component-oriented engineering for intelligent designs, as it preserves the behaviours of the original controllers while introducing additional functionality. To demonstrate the efficacy of our approach we utilise an existing autonomous driving case study as well as a set of smaller benchmarks. These show that introduced overheads are extremely minimal and that the approach is very scalable.

cs.SE↗

arXiv2021

FLAW3D: A Trojan-based Cyber Attack on the Physical Outcomes of Additive Manufacturing

Additive Manufacturing (AM) systems such as 3D printers use inexpensive microcontrollers that rarely feature cybersecurity defenses. This is a risk, especially given the rising threat landscape within the larger digital manufacturing domain. In this work we demonstrate this risk by presenting the design and study of a malicious Trojan (the FLAW3D bootloader) for AVR-based Marlin-compatible 3D printers (>100 commercial models). We show that the Trojan can hide from programming tools, and even within tight design constraints (less than 1.7 kilobytes in size), it can compromise the quality of additively manufactured prints and reduce tensile strengths by up to 50%.

cs.CR↗

arXiv2020

DAVE: Deriving Automatically Verilog from English

While specifications for digital systems are provided in natural language, engineers undertake significant efforts to translate them into the programming languages understood by compilers for digital systems. Automating this process allows designers to work with the language in which they are most comfortable --the original natural language -- and focus instead on other downstream design challenges. We explore the use of state-of-the-art machine learning (ML) to automatically derive Verilog snippets from English via fine-tuning GPT-2, a natural language ML system. We describe our approach for producing a suitable dataset of novice-level digital design tasks and provide a detailed exploration of GPT-2, finding encouraging translation performance across our task sets (94.8% correct), with the ability to handle both simple and abstract design tasks.

cs.SE↗

arXiv2020

Designing Neural Networks for Real-Time Systems

Artificial Neural Networks (ANNs) are increasingly being used within safety-critical Cyber-Physical Systems (CPSs). They are often co-located with traditional embedded software, and may perform advisory or control-based roles. It is important to validate both the timing and functional correctness of these systems. However, most approaches in the literature consider guaranteeing only the functionality of ANN based controllers. This issue stems largely from the implementation strategies used within common neural network frameworks -- their underlying source code is often simply unsuitable for formal techniques such as static timing analysis. As a result, developers of safety-critical CPS must rely on informal techniques such as measurement based approaches to prove correctness, techniques that provide weak guarantees at best. In this work we address this challenge. We propose a design pipeline whereby neural networks trained using the popular deep learning framework Keras are compiled to functionally equivalent C code. This C code is restricted to simple constructs that may be analysed by existing static timing analysis tools. As a result, if compiled to a suitable time-predictable platform all execution bounds may be statically derived. To demonstrate the benefits of our approach we execute an ANN trained to drive an autonomous vehicle around a race track. We compile the ANN to the Patmos time-predictable controller, and show that we can derive worst case execution timings.

cs.LG↗