Confabulators, Persuaders, and Imagination Catalysts: An Expert Perspective on the Chatbot Phenomenon

In all my years as a researcher, I’ve never had so many friends and family members asking me about AI – chatbots, in particular. Even people that I would have described as fairly disinterested in tech in general have shared with me their experiences interacting with ChatGPT, or expressed that they are fearful and/or intrigued […]

Read More

Proof Assistance and Repair in Crux

Overview We have added support for semi-automated proof assistance and repair to Crux, Galois’s symbolic testing tool for C/C++ and Rust. These new capabilities build on support for logical abduction provided by the cvc5 SMT solver that suggests possible facts for failed proof goals, that, when assumed, make the proof goals provable. This feature can […]

Read More

Generative AI, Mission Critical Systems, and the Tightrope of Trust

The public release of ChatGPT3 and DALL-E 2 radically changed our expectations for the near future of AI technologies. Given the demonstrated capability of large generative models (LGMs), the ways in which they immediately captured public imagination, and the level of publicized planned capital investment, we can anticipate rapid integration of these models into current […]

Read More

Harnessing Deep Learning to Model Complex Systems

Researchers at Galois have developed DLKoopman – an open-source software tool that uses machine learning to model and predict the behavior of complex, difficult-to-analyze systems. DLKoopman models a system from limited data, and then predicts how it is going to behave under unknown, often unmeasurable conditions, such as the pressure on a submarine at unknown […]

Read More

2022: Year in Review

2022 wasn’t a return to pre-pandemic times, but we’ll call it a “return to almost-normal.” At Galois, we continued a hybrid model of in-person and virtual collaboration. We also added several members to our team, including the acquisition of Adventium Labs! (More on that below.) Throughout the year, we made considerable progress applying our efforts […]

Read More

Identifying Design Choices That Increase a System’s Exploitability

We know that software flaws can create vulnerabilities within a system. But a system’s design often impacts whether a software flaw can be exploited. For example, subtle differences between a system’s ideal design and its implementation can lead to different emergent run-time behaviors. These emergent behaviors can act like a “programmable weird machine” – making […]

Read More

Driving to a Secure Future: Demonstrating a Vehicle That Thwarts Cyberattacks

The National Cryptologic Museum opened its doors to the public last week. As part of the exhibits, visitors will be able to interact with a quirky little car with a big claim: under the hood, it demonstrates hardware that can thwart many cyberattacks on automobiles. The BESSPIN Vehicle Demonstrator DARPA’s System Security Integration Through Hardware […]

Read More