Recent advances in deep learning (DL) allow for solving complex AI problems that used to be considered very hard. While this progress has advanced many fields, it is considered to be bad news for Completely Automated Public Turing tests to tell Computers and Humans Apart (CAPTCHAs), the security of which rests on the hardness of some learning problems. In this paper, we introduce DeepCAPTCHA, a new and secure CAPTCHA scheme based on adversarial examples, an inherit limitation of the current DL networks. These adversarial examples are constructed inputs, either synthesized from scratch or computed by adding a small and specific perturbation called adversarial noise to correctly classified items, causing the targeted DL network to misclassify them. We show that plain adversarial noise is insufficient to achieve secure CAPTCHA schemes, which leads us to introduce immutable adversarial noise-an adversarial noise that is resistant to removal attempts. In this paper, we implement a proof of concept system, and its analysis shows that the scheme offers high security and good usability compared with the best previously existing CAPTCHAs.
In this work, we developed an automated system to detect potentially illegal elephant ivory items for sale on eBay. Two law enforcement experts, with specific knowledge of elephant ivory identification, manually classified items on sale in the Antiques section of eBay UK over an 8 week period. This set the “Gold Standard” that we aim to emulate using data-mining. We achieved close to 93% accuracy with less data than the experts, as we relied entirely on metadata, but did not employ item descriptions or associated images, thus proving the potential and generality of our approach. The reported accuracy may be improved with the addition of text mining techniques for the analysis of the item description, and by applying image classification for the detection of Schreger lines, indicative of elephant ivory. However, any solution relying on images or text description could not be employed on other wildlife illegal markets where pictures can be missing or misleading and text absent (e.g., Instagram). In our setting, we gave human experts all available information while only using minimal information for our analysis. Despite this, we succeeded at achieving a very high accuracy. This work is an important first step in speeding up the laborious, tedious and expensive task of expert discovery of illegal trade over the internet. It will also allow for faster reporting to law enforcement and better accountability. We hope this will also contribute to reducing poaching, by making this illegal trade harder and riskier for those involved.
Existence of mobile devices with high performance cameras and powerful image processing applications eases the alteration of digital images for malicious purposes. This work presents a new approach to detect digital image tamper detection technique based on CFA artifacts arising from the differences in the distribution of acquired and interpolated pixels. The experimental evidence supports the capabilities of the proposed method for detecting a broad range of manipulations, e.g., copy-move, resizing, rotation, filtering and colorization. This technique exhibits tampered areas by computing the probability of each pixel of being interpolated and then applying the DCT on small blocks of the probability map. The value of the coefficient for the highest frequency on each block is used to decide whether the analyzed region has been tampered or not. The results shown here were obtained from tests made on a publicly available dataset of tampered images for forensic analysis. Affected zones are clearly highlighted if the method detects CFA inconsistencies. The analysis can be considered successful if the modified zone, or an important part of it, is accurately detected. By analizing a publicly available dataset with images modified with different methods we reach an 86% of accuracy, which provides a good result for a method that does not require previous training.
Widespread adoption of RFID technology is being slowed down because of increasing public concerns about associated security threats. This paper shows that it is possible to enhance the security of RFID systems by requiring readers to perform a computational effort test. Readers must solve a cryptographic puzzle - one of the components of the Weakly Secret Bit Commitment (WSBC) sent by tags - to obtain the static identifier of the interrogated tag. The method we present is based on a simple concept already used in security applications such as anti-spam or TCP SYN flooding protection, yet original in the RFID context until now. The scheme provides privacy protection while being an effective countermeasure against the indiscriminate disclosure of the whole contents of a large number of tags. Then, we scrutinize the combined use of cryptographic puzzles and distance-bounding protocols. First, a classical and relatively straight-forward solution is presented. Secondly, we introduce a protocol named Noent, that follows a new approach that reduces drawbacks associated with WSBC such as key delegation, whilst gaining all the advantages of employing distance-bounding protocols such as the certainty on the distance between a tag and reader.
The prevalence and significance of web services in our daily lives make it imperative to ensure that they are – as much as possible – free from vulnerabilities. However, developing a complex piece of software free from any security vulnerabilities is hard, if not impossible. One way to progress towards achieving this holy grail is by using static code analysis tools to root out any common or known vulnerabilities that may accidentally be introduced during the development process. Static code analysis tools have significantly contributed to addressing the problem above, but are imperfect. It is conceivable that static code analysis can be improved by using AI-powered tools, which have recently increased in popularity. However, there is still very little work in analysing both types of tools’ effectiveness, and this is a research gap that our paper aims to fill. We carried out a study involving 11 static code analysers, and one AI-powered chatbot named ChatGPT, to assess their effectiveness in detecting 92 vulnerabilities representing the top 10 known vulnerability categories in web applications, as classified by OWASP. We particularly focused on PHP vulnerabilities since it is one of the most widely used languages in web applications. However, it has few security mechanisms to help its software developers. We found that the success rate of ChatGPT in terms of finding security vulnerabilities in PHP is around 62-68%. At the same time, the best traditional static code analyser tested has a success rate of 32%. Even combining several traditional static code analysers (with the best features on certain aspects of detection) would only achieve a rate of 53%, which is still significantly lower than ChatGPT’s success rate. Nonetheless, ChatGPT has a very high false positive rate of 91%. In comparison, the worst false positive rate of any traditional static code analyser is 82%. These findings highlight the promising potential of ChatGPT for improving the static code analysis process but reveal certain caveats (especially regarding accuracy) in its current state. Our findings suggest that one interesting possibility to explore in future works would be to pick the best of both worlds by combining traditional static code analysers with ChatGPT to find security vulnerabilities more effectively.