The Human Centred Computing research group in Computer Science here in Oxford investigates the ways in which technology affects communication, collaboration and knowledge exchange within scientific, work and home settings, and considers how new technologies can be designed and developed to be more responsive to societal needs. The group is one of the leading centres of expertise in Responsible Innovation, an emerging set of practices which aim to ensure a proper embedding of scientific and technological advances in society. Responsible Innovation recognises that we cannot predict the future but we can try to anticipate plausible outcomes, so that we can be prepared for the social changes brought by new technologies. Rather than limiting innovation or being another "hoop to jump through", Responsible Innovation provides a creative resource to understand technology in new ways. This talk will discuss the promises and challenges raised by our Responsible Innovation investigations into quantum computing and related technologies. Inter-disciplinary researchers (physicists, engineers, materials scientists, computer scientists) at Oxford are leading one of the major efforts to build a quantum computer in the Networked Quantum Information Technologies (NQIT) Hub. Quantum technologies have important applications for defence and security. We already know that a powerful quantum computer will be able to break existing Internet encryption – but a sufficiently large computer is still years away. Conversely, quantum cryptography, which is already being offered as a commercial service, uses the principles of quantum mechanics to provide information-theoretic security which cannot be eavesdropped. Powerful technologies such as these raise questions: who will have access to this technology? What other applications of will appear, as technologies become reality and are widely adopted? How may they be used for the benefit of society? I will draw on the insights of Responsible Innovation to explore these questions in the context of quantum computing. If there is time, I will also discuss some other very exciting work of the group.

Cyber Security Oxford and the Centre for Technology and Global Affairs are delighted to host Thomas Dullien/Halvar Flake from Google's Project Zero in our cyber security seminar series. He will give a talk about his most recent work on the formalisation of exploitation. If you are interested in meeting Thomas during his time at Oxford, please contact florian.egloff@pmb.ox.ac.uk. Abstract In spite of being central to everything that is going on in IT security, the concept of "exploit" is surprisingly poorly formalized and understood only on an intuitive level by security practitioners. This lack of clear definition has all sorts of negative side-effects: From ineffictive teaching to muddled thinking about mitigations. In this talk, I will make an attempt to more clearly define what it is that attackers do when they write an exploit – and then talk about what this means for mitigations and secure coding. The concept of exploit is central to computer security, particularly in the context of memory corruptions. Yet, in spite of the centrality of the concept and voluminous descriptions of various exploitation techniques or countermeasures, a good theoretical framework for describing and reasoning about exploitation has not yet been put forward. A body of concepts and folk theorems exists in the community of exploitation practitioners; unfortunately, these concepts are rarely written down or made sufficiently precise for people outside of this community to benefit from them. This paper clarifies a number of these concepts, provides a clear definition of exploit, a clear definition of the concept of a weird machine, and how programming of a weird machine leads to exploitation. The paper also shows, somewhat counterintuitively, that it is feasible to design some software in a way that even powerful attackers - with the ability to corrupt memory once - cannot gain an advantage. The approach in this paper is focused on memory corruptions. While it can be applied to many security vulnerabilities introduced by other programming mistakes, it does not address side channel attacks, protocol weaknesses, or security problems that are present by design.

All kinds of previously local services are being moved to a cloud setting. While this is justified by the scalability and efficiency benefits of cloud-based services, it also raises new security and privacy challenges. Solving them by naive application of standard security/privacy techniques can conflict with other functional requirements. In this talk, I will outline some cloud-assisted services and the apparent conflicts that arise while trying to secure these services. Taking the case of cloud-assisted malware scanning as an example scenario, I will discuss the privacy concerns that arise and how we can address them effectively. I will then discuss a more general setting of using cloud-hosted machine learning models in a privacy-preserving manner.

Abstract: The WannaCry and (Not)Petya attacks have generated concern over the speed and impact of the dissemination of state hacking-tools into the public domain. Yet, their public release means these attacks did not utilise 0-day exploits – with the vulnerabilities patched for the most common operating systems. Due to environments, or ecologies, impacts are emergent, cascading and complex according to the spatial complexities of cyberspace.  Previous academic study concerning malware has extensively considered the ‘virtual’ or ‘cyber’ weapon, or focused on cybercriminals. This talk wishes to broaden this horizon – through complicating and complementing current theorisation, to include what a geography of malware tells us. Drawing on (auto)ethnographic work of a malware analysis laboratory, this talk considers the use of ecology to explore the implications of moving away from pathology (that of signatures and static renderings of code) through to limited attempts to consider broader context in machine learning. The inspiration for ecology (or ecosophy) draws on the philosophical work of the French philosopher, Félix Guattari, in which he draws together the triad of: social relations; environment; and human subjectivity. Using this thinking lets us explore how (malicious) software becomes operable, performs its existence, and how conventional distinctions between different, apparently separate elements are threaded together that lead to malware being ‘successful’.  Working in this way provides an extensive analysis of how geography and IR can inform one another to explore the contours of what can thus be considered malicious.

Defending against botnets has always been a cat and mouse game. Cyber-security researchers and government agencies attempt to detect and take down botnets by playing the role of the cat. In this context, a lot of work has been done towards reverse engineering certain variants of malware families as well as understanding the network protocols of botnets to identify their weaknesses (if any) and exploit them. While this is necessary, such an approach offers the botmasters the ability to quickly counteract the defenders by simply performing small changes in their arsenals. We attempt a different approach by actually taking the role of the Botmaster, to eventually anticipate his behavior. That said, in this presentation, we present a novel computational trust mechanism for fully distributed botnets that allows for a resilient and stealthy management of the infected machines (zombies). We exploit the highly researched area of computational trust to create an autonomous mechanism that ensures the avoidance of common botnet tracking mechanisms such as sensors and crawlers. In our futuristic botnet, zombies are both smart and cautious. They are cautious in the sense that they are careful with whom they communicate with. Moreover, they are smart enough to learn from their experiences and infer whether their fellow zombies are indeed who they claim to be and not government agencies' spies. We study different computational trust models, mainly based on Bayesian inference, to evaluate their advantages and disadvantages in the context of a distributed botnet. Furthermore, we show, via our experimental results, that our approach is significantly stronger than any technique that has been seen in botnets to date. Finally, we step out of the adversarial perspective and touch the topic of countermeasures against our own approach.

The Global Cyber Security Capacity Centre had started to build a network of partners around the world to achieve maximum leverage for the Centre’s work internationally. Pursuing a regional strategy it has started to establish regional collaboration partnerships with centres of excellence in key locations around the world. The first collaboration has been established with the Oceania Cyber Security Centre (OCSC), Melbourne, which will seek to deploy the CMM in the 15 countries of the Oceania region by 2019. The team from the OCSC, which is currently in Oxford, will present an overview of the cybersecurity landscape in the Oceania region and what they aim to identify in the countries with the deployment of CMM.

PROTECTIVE is a cyber threat intelligence sharing platform being developed by a consortium of ten partners from eight European countries, including: three European National Research Educational Networks (NRENs – CESNET, PSNC and RoEduNet), three academic (Oxford, Athlone and Darmstadt) and four commercial partners (ITTI, GMV, Email Laundry and Synyo). The platform is a suite of threat intelligence sharing tools that aim at providing security teams with a greater context, threat and situational awareness, and thus improve an organisation's ongoing awareness of risks posed to it by attacks. Specifically, the platform is designed to provide solutions for public domain CSIRTs outside the mainstream of cyber security solution provision. Public CSIRTs needs arise in part because commercial tools do not address their unique requirements. This has created a shortfall, clearly articulated by ENISA, of tools with the required analytical and visualization capabilities to enable public CSIRTs to provide optimised services to their constituencies. In the PROTECTIVE project we investigate the state of the art in cyber threat intelligence generation and sharing, and are developing a solution to: enhance security alert correlation and prioritisation, link the relevance and criticality of an organisations assets to its business/mission, investigate how computational trust can be used in threat intelligence sharing, and finally establish a public CSIRT threat intelligence sharing community to leverage these solutions. In order to do so, we are studying the current technological and human factor challenges, and attempt to identify what has limited threat intelligence sharing tools from flourishing in the past. In this talk, I will outline project activities to date, and present our lessons learnt from the first year of research and development. I will also outline the key new capabilities of the platform, related to: threat intelligence aggregation, enrichment, sharing automation, community creation, trust computation (confidence in quality of the data, as opposed to trust in the transportation layer), and General Data Protection Regulation (GDPR) compliance.

Mini-abstract: This article examines concepts of cybersecurity (im)maturity in the states of the Gulf Cooperation Council (GCC), drawing on interviews with cybersecurity professionals in the region. It distinguishes between two aspects of immaturity. First, tickbox immaturity is the implementation of standards in a ‘deficient’ manner, through surface-level implementation of prescriptions rather than a deeper attitude of proactive risk management. Second, blackbox immaturity is the reliance on technologies manufactured elsewhere, which are often “shiny toys”: high level, sophisticated defence systems inappropriate to the real threats faced.Abstract: In International Relations (IR), the concept of maturity echoes earlier ideas such as civilization and development, which structured the global order in explicitly unequal ways as part of colonial and postcolonial relationships between the global centre – usually the ‘West’ - and its various peripheries. Despite these connotations of historic exploitation, a ‘vanilla’ version of maturity is common in international security practices, largely drawn from the discourse of international consultants, organizational theory, and the applied sciences. This article uses critical and postcolonial theory in IR and science and technology studies (STS) to investigate maturity, and argues that current concepts of maturity are vulnerable to similar critiques as those levelled against earlier universals.This article uses the case study of cybersecurity in the states of the Gulf Cooperation Council (GCC): Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and the UAE. It draws on 30 qualitative semi-structured interviews with cybersecurity professionals in these countries conducted in 2016 and 2017, as well as a range of other sources. It distinguishes between two aspects of immaturity, which I call ‘tickbox’ and ‘blackbox’. Tickbox immaturity is the implementation of standards in a ‘deficient’ manner, through surface-level implementation of prescriptions rather than a deeper attitude of proactive risk management. Blackbox immaturity is the reliance on technologies manufactured elsewhere, as part of outsourcing and maintenance contracts by external parties. Both aspects of maturity are identified by individuals in all three key subject positions in GCC cybersecurity: Western contractors, local managers, and expatriate employees. However, they locate the sources of tickbox immaturity differently: some point to essentialist notions of culture, and others to unequal power dynamics around immigration. Furthermore, I argue that blackbox immaturity is a result of commercial struggles by Western companies to define what counts as cybersecurity expertise. This creates an Orientalized view of cybersecurity in the region as being centred only on “shiny toys”: high level, sophisticated defence systems which are often inappropriate for the real threats faced.

Modern architectures such as the blockchain can enable distributed parties to achieve common knowledge and agreed timestamping.  We exploit this to develop a new and extremely efficient form of signature which has the advantage of almost certainly being quantum resistant.  We provide definitions of signature formation and verification, PKI and certification authorities (CA) including a form of zero-knowledge proof by which Alice can convice a CA that she knows her "private key".

Title:  System Intrusion Tolerance   Speaker:  Mike Fisk, Los Alamos National Laboratory A great deal of effort is spent trying to prevent cyber intrusions and yet commodity systems continue to be penetrable.   Even when software is kept patched, social engineering, zero-day vulnerabilities, and supply chain risks continue to provide would-be intruders with ample opportunities.   Our work is geared at maintaining some degree of system security after a system has been compromised.   We will present an overview of our work on anomalous change detection within networks, and on quantitative measures of intrusion tolerance for authentication technologies.     Title:  Balancing Privacy, Organizational Risk, and Information Sharing   Speaker:  Gina Fisk, Los Alamos National Laboratory Sharing cybersecurity data across organizational boundaries brings both privacy risks in the exposure of personal information and data, and organizational risk in disclosing internal information.   These risks occur as information leaks in network traffic or logs, and also in queries made across organizations.  They are also complicated by trade-offs in privacy preservation and utility present in anonymization to manage disclosure.  In this presentation we will present three principles that we use to guide sharing security information across organizations, and then will discuss engineering techniques that apply these principles to a distributed security system.  Application of these principles can reduce the risk of data exposure and help manage trust requirements for data sharing, helping to meet our goals of balancing privacy, organizational risk, and the ability to better respond to security threats with shared information.