Cybercriminal activities have increased dramatically over the last few years, especially as a result of both the technological developments and the perceived anonymity of users of such technologies. Common users often engage in criminal activities online (such as intellectual property theft, stalking, defamation, cyberbullying, cyberthreats and terrorism, among others) as they believe that, by creating fake profiles and using public access computers, their identity remains unknown; computer expert users, on the contrary, have the knowledge and skills to obfuscate their real identity, e.g. by using stealth technologies or even hard core methods such as the ones provided by the dark web. Forensic Linguistics, and especially its subarea of Forensic Authorship Analysis, has been crucial in this area as it allows the identification of the most likely author from a pool of suspects by analysing the language used. As previous research shows, traditional forensic authorship analysis methods and techniques have been successfully used in online and offline criminal cases alike. However, the future of computing will raise new challenges to the forensic analysis. For instance, the fact that AI systems such as the OpenAI are able to generate human-like texts, by supposedly emulating an authors’ writing style, can turn out to become the single biggest challenge for forensic linguists, as research conducted so far has demonstrated that humans are incapable of replicating another author’s writing style — their idiolect. Notwithstanding, such systems can also be used to aid forensic linguists, as long as the latter can gain the knowledge required to efficiently use systems available. This session discusses how present computing is threatening forensic work, and how such threats will increase in the future, to argue that, ultimately, the power to counter ‘black-box’ systems – as required by law – is in the hands of future forensic linguistic computing.
Friday, 5 July 2019
Future of Computing, Porto, Portugal