A new industry-academic partnership led by the University of Toronto will accelerate the development and commercialization of next-generation electric vehicle (EV) and mobility technologies.

In collaboration with seven other southern Ontario universities, Electric Vehicle Innovation Ontario (EVIO) will embed 37 highly skilled graduate researchers directly inside 20 Ontario EV and mobility companies.

Today, the Federal Economic Development Agency for Southern Ontario announced a $2.5-million contribution to EVIO. Matched by industry and academic partners for a total program value of $7.9 million, EVIO is expected to generate over $30 million in economic activity, expand firm-level research and development capacity and accelerate the creation of new Canadian intellectual property in EV.

“AI and clean technology are vital to helping build a strong economic future for Canada,” said the Honourable Evan Solomon, Minister of Artificial Intelligence and Digital Innovation and Minister responsible for the Federal Economic Development Agency for Southern Ontario. “Through this investment in University of Toronto’s Electric Vehicle Innovation Ontario, we are backing Canadian ingenuity to grow a world-class EV supply chain, strengthen our competitive advantage, and create good, meaningful jobs. This is how we build Canada strong: with innovation, skills and a clean economy that works for everyone.”

The researchers will work on real-world challenges in battery chemistry, charging reliability, power electronics, mobility software, cold-weather performance and advanced manufacturing — areas critical to promoting EV adoption and strengthening Canada’s position in a rapidly evolving global EV market.

“EVIO connects graduate researchers directly with industry, speeding up the development of advanced EV technologies while generating new IP and future economic growth for Canada,” said Arvind Gupta, professor in U of T’s ������Ƶ of Computer Science and scientific director of EVIO. “This is exactly the kind of partnership that positions Canadian innovators to lead globally.”

Industry partners contribute $45,000 towards a $90,000 project designed to enable companies to scale innovations while providing researchers with competitive compensation, hands-on experience and direct pathways into high-growth careers.

“EVIO is a strong example of what Canada can achieve when industry, academia, and government work together with purpose,” said Paul Slaby, managing director of Canada’s Semiconductor Council. Slaby noted that programs like EVIO help move ideas into the marketplace while strengthening the talent pipeline.

Beyond U of T, EVIO’s network of participating universities includes Queen’s University, Toronto Metropolitan University, University of Ottawa, University of Waterloo, University of Windsor, Western University and York University.

“EVIO represents the kind of industry-research partnership Canada needs to meet our ambitions: agile, market-driven, and anchored in scientific excellence,” said Charmaine Dean, vice-president of research and international, and professor in the ������Ƶ of Statistics and Actuarial Science at the University of Waterloo and chair of the EVIO Steering Committee. “By embedding top researchers directly inside firms, we are closing the gap between discovery and deployment. This is how we accelerate commercialization, strengthen Ontario’s EV ecosystem and ensure Canada remains a competitive force in the global economy.”

EVIO is modeled on internationally proven approaches, including Germany’s Fraunhofer Institutes and DARPA-style applied innovation teams, which have catalyzed breakthrough technologies across G7 economies for decades.

Interested in getting involved?

If you’re an EV or mobility company looking to advance your innovation agenda, connect with us today: evio.dcs@utoronto.ca

What began as a small event designed for Master of Science in Applied Computing (MScAC) students 10 years ago has grown into one of Toronto’s largest academic-led tech innovation showcases. The annual Applied Research in Action (ARIA) showcase now highlights cutting-edge research from across the University of Toronto’s ������Ƶ of Computer Science.

More than 1,000 students, alumni, faculty, staff, industry leaders and government officials attended this year’s event, which showcased the department’s growing impact in applied research.

“The rapid growth of ARIA attests to the importance of the event for Toronto’s tech space,” said Arvind Gupta, professor, academic director of professional programs and director for strategic initiatives in the ������Ƶ of Computer Science. “Many companies come out to see for themselves the amazing technologies being developed. This inspires even more companies to develop next-generation technologies.”

Sponsors included AMD, Mitacs, Vanguard, Cresta, Georgian, Geotab, Shopify, Qorsa, Next Pathway and Ubisoft. These partnerships go beyond sponsorship, Gupta said; they represent a shared commitment to mentorship, meaningful engagement and real-world impact.

The event also featured the MScAC Award Ceremony. Special guests MP , MP and CEO presented the Industry Icon Award, the Student Innovation Award and the new MScAC-Mitacs Award.

Award recipients:

• Industry Icon Award: Jithin Pradeep, director, head of enterprise AI & research, Vanguard

• Student Innovation Award: Frank Bai, MScAC student

• Faculty Recognition Award: Xiaofei Shi, assistant professor, ������Ƶ of Statistical Sciences

• ARIA Spotlight Award: Alan Rosenthal, system administrator, MScAC

• MScAC-Mitacs Award: Ashka Shah, MScAC student

After the ceremony, Stephen Lucas of Mitacs, Deidre Haskell, director of the , and Eyal de Lara, professor and chair of the ������Ƶ of Computer Science, signed a partnership agreement to raise awareness of Canadian computer science graduate programs in India.

ARIA 2025 featured an impressive 140 student-led projects from undergraduate, MScAC, MSc and PhD researchers, many developed in collaboration with industry partners in sectors such as AI, health, finance and cybersecurity. Attendees explored emerging technologies, engaged directly with student researchers and learned innovative solutions to real-world challenges.

Across all levels, students and researchers are applying their skills to meaningful, industry-relevant problems. Three projects stood out for their use of emerging technologies — these stories highlight how computer science researchers are making an impact.

From FTL to ARIA: students bridging leadership and research

The is an experiential learning program that prepares undergraduate students to lead in technology-driven industries. Industry partners present problems in class, and student teams design software solutions and business plans.

Carlos Solares, Tara Tandon and their team showcased their FTL project at ARIA 2025. Their work addresses a major fintech challenge: reducing bias in machine learning systems used for fraud detection.

In collaboration with Cash App, the team built a platform that helps executives see how bias in fraud detection affects users and business outcomes. The goal is to give companies the tools to spot and reduce unfair patterns, especially for groups that may be disproportionately targeted.

“Presenting at such a large, high-profile event was exhilarating and an honour,” said Solares and Tandon. “It will no doubt be a highlight of our undergraduate journeys.”

They credit the FTL program with providing industry exposure and helping them develop leadership, networking, presentation and collaboration skills.

“Being surrounded by hard-working peers and given the resources and guidance to grow our skills has led many of us to land internships, pursue leadership roles and present research at conferences like ARIA.”

Innovating for impact: student’s research highlights the future of cancer care

Showcasing the impact of student-driven innovation, Ashka Shah presented at ARIA, detailing how her eight-month applied research internship at the advanced the use of data-science techniques to detect lymphoma from routine, non-invasive procedures.

Routine imaging tests such as PET scans can sometimes miss early signs of cancer. But biology offers another way to detect cancer. As Shah explains, tumour cells release tiny amounts of DNA into the bloodstream. Her research explores how these biological signals could lead to more sensitive and accurate detection methods.

Inspired by family tragedy, Shah focuses on identifying signatures that distinguish tumour-derived DNA fragments from the background in liquid biopsies, or blood draws.

"My grandfather was diagnosed with cancer at a late stage and passed away,” said Shah. “Things might have been different if we’d had sensitive, accessible and non-invasive early cancer detection techniques.”

On the day of ARIA, she discussed her work in a live radio interview on (at 21:43), calling the experience “a dream come true.”

“I never imagined explaining my project live on air to thousands of listeners,” she said. “My goal was to introduce the concept of liquid biopsies for cancer detection. I'd consider it a success if even one person felt hopeful about the future of cancer diagnostics.”

She said connecting with professionals who share her vision of advancing health tech was inspiring. “It gives us a chance to showcase our projects beyond academia to industry leaders who can help turn proofs of concept into real-world solutions.”

Looking ahead, Shah said her research is in its early stages — a proof-of-concept with a small cohort. Clinical applications will need further validation, but the potential impact is significant.

Researchers expose GPU flaw that threatens AI reliability

The PhD program trains researchers to tackle major challenges in computing and AI. One project featured at ARIA — GPUHammer — revealed a critical security flaw in widely used graphics processing units (GPUs).

GPUs power AI systems, video games and data analysis. U of T researchers showed these chips are vulnerable to Rowhammer, an attack that repeatedly accesses specific memory locations to corrupt nearby data. This means someone using a shared GPU could potentially corrupt information or interfere with how an AI system works.

Assistant Professor , second-year PhD student Chris (Shaopeng) Lin and fourth-year undergrad Joyce Qu developed the proof-of-concept attack against GDDR6 memory on an NVIDIA RTX A6000, a GPU widely used for high-performance computing.

“More investigation will probably reveal more issues,” says Saileshwar. “And that’s important, because we’re running incredibly valuable workloads on GPUs. AI models are being used in real-world settings such as healthcare, finance, and cybersecurity. If there are vulnerabilities that allow attackers to tamper with those models at the hardware level, we need to find them before they’re exploited.”

ARIA 2025 showed the ������Ƶ of Computer Science is a hub for transformative ideas, where students and researchers at every level apply emerging technologies to real-world challenges. The projects highlighted trends shaping the future of tech, from responsible AI and cybersecurity to health innovation and large-scale data systems.

Industry exhibitors spanning sectors such as finance, gaming, health and enterprise AI showed how academic research and industry needs increasingly intersect. These collaborations underscored both the depth and diversity of work across the department and reaffirmed U of T’s role in shaping Toronto’s tech landscape and beyond.

University of Toronto theoretical computer scientist has been awarded the for his pioneering contributions that have impacted algorithmic challenges underlying modern society.

Presented annually by the Infosys Science Foundation across six categories, the prize recognizes outstanding early-career researchers and scientists of Indian origin or whose work impacts India. It comes with a gold medal, formal citation and US$100,000.

In his research, Sachdeva focuses on the design of fast algorithms for graph problems, as well as mathematical machine learning.

The prize citation highlights Sachdeva’s “deep insights into mathematical optimization and the resolution of longstanding open questions in algorithmic theory.”

“His work has established new standards on achievable performance in computational problems affecting information flows across societal lifelines, including the internet, transportation and communication networks,” the citation reads.

Sachdeva is an associate professor in the ������Ƶ of Mathematical and Computational Sciences at the University of Toronto Mississauga and the tri-campus graduate ������Ƶ of Computer Science.

He has been recognized by fellow researchers for developing “absurdly fast” algorithms that have created true breakthroughs in the foundational methods of computer science — a rare occurrence for computational challenges that have persisted for decades.

Jury chair Jayathri Murthy notes Sachdeva’s work “will influence fields as diverse as healthcare, transportation and education for years to come.”

The ������Ƶ of Computer Science at the University of Toronto, and the have announced a new collaboration to build stronger research ties with some of the best universities in India. Leveraging the success of Mitacs Globalink Research Internship program and the Fields Undergraduate Summer Research Program, the three organizations have joined forces to establish the Fields-Mitacs Summer Research Program.    

The Fields-Mitacs Summer Research Program will bring top senior undergraduate students from select institutions in India to participate simultaneously in the Mitacs Globalink Research Internship program and the Fields Undergraduate Summer Research Program. The University of Toronto ������Ƶ of Computer Science is recruiting professors to put forward projects and host students, as well as soliciting participation from top institutions in India. 

Students will be invited by their home institutions to apply to the program, with nominated students undergoing a matching process in early 2026. The first cohort of students will come to Canada during Summer 2026.  

This new initiative will provide increased visibility for higher education institutions in Southern Ontario and attract top talent to contribute to research projects under the supervision of Canadian faculty. Program participants will receive mentorship, training and exposure to future opportunities for industry collaboration.

Up to 12 students will participate in the initial cohort, with students coming from select partner institutions in India. 

Quotes

About the University of Toronto ������Ƶ of Computer Science  

The ������Ƶ of Computer Science at the University of Toronto is a global leader in computing research and education, consistently ranked among the top computer science departments worldwide. Known for its pioneering contributions to areas such as artificial intelligence, machine learning, human-computer interaction and systems, the department fosters a vibrant academic community that brings together world-renowned faculty, innovative researchers, and ambitious students. With strong ties to industry and a collaborative, interdisciplinary approach, U of T’s ������Ƶ of Computer Science drives technological advancement and prepares graduates to become leaders in academia, industry and beyond.   

The University of Toronto, founded in 1827, is Canada’s leading public research university, recognized globally for its academic excellence and innovation. Across its three campuses, the university offers over 700 undergraduate and 200 graduate programs. U of T is home to world-changing discoveries and continues to shape a better future through research, teaching and public impact. 

About Mitacs 

For over 25 years, Mitacs has helped grow the economy and develop the workforce of tomorrow, connecting industry with academia and global partners to solve real-world challenges. We support business-academic research collaboration through internships, co-funded with businesses, for undergraduate to graduate students and post-doctoral fellows. As a national innovation connector, Mitacs takes a talent-first approach to strengthen innovation capacity and drive global competitiveness. We serve as an essential research-commercialization bridge, accelerating market entry and growth for new products and services. This is a critical time for Canada to think big and take bold action. Mitacs is ready to help build a strong and resilient Canadian economy, powered by ideas, talent and innovation.

Mitacs is funded by the Government of Canada, the Government of Alberta, the Government of British Columbia, Research Manitoba, the Government of New Brunswick, the Government of Newfoundland and Labrador, the Government of Nova Scotia, the Government of Ontario, Innovation PEI, the Government of Quebec, the Government of Saskatchewan, and the Government of Yukon. 

About the Fields Institute 

At the Fields Institute, mathematics research, innovation and education flourish. We foster an inclusive, equitable and collaborative culture where everyone can discover mathematics, and where mathematicians can make meaningful contributions to the world. Our mission is simple: we seek to bring together mathematicians from across sectors and the globe to develop the mathematics that will shape our future. 

The Fields Institute is primarily funded by the Federal and Provincial governments, nine Principal Sponsoring Universities from across Ontario, and 15 Affiliate Sponsoring Universities. We are also supported by numerous other Institutions and Centres across Canada and internationally. 

Renowned artificial intelligence researchers and University of Toronto computer science professors and have been by Schmidt Sciences. 

The AI2050 program supports researchers whose projects use artificial intelligence to create ‘immense benefits’ for humanity by 2050. This year’s 21 early career fellows and seven senior fellows are eligible to receive more than $18 million in fellowships. 

Alán Aspuru-Guzik, a professor jointly appointed to the U of T departments of chemistry and computer science, conducts research in the interfaces of quantum information, machine learning and chemistry. 

He is the Canada 150 Laureate in Theoretical Chemistry and a Canada CIFAR AI Chair at the Vector Institute. He is also a CIFAR Fellow co-directing the Accelerated Decarbonization program. 

Apuru-Guzik leads U of T’s , a strategic initiative that gathers researchers from industry, government and academia to envision and create the lab of the future. 

His AI2050 project involves developing an “AI chemist," an artificial intelligence system designed to work alongside human chemists to accelerate scientific discoveries in chemistry. 

Sheila McIlraith is a professor in the ������Ƶ of Computer Science and an associate director and research lead for the at U of T. She is also a Canada CIFAR AI Chair at the Vector Institute. Her research focuses on AI sequential decision making through the lens of human-compatible AI. 

McIlraith co-leads U of T’s , which teaches students how to incorporate ethical considerations into the design and deployment of technology. 

In her AI2050 project, McIlraith aims to endow AI with purposeful theory of mind capabilities — the ability not only to better understand the beliefs, desires, and intentions of others, but also to be incentivized to make decisions and to act in consideration of the welfare and agency of others.

“The AI2050 fellows are ambitious yet collaborative researchers who focus on AI innovation and the opportunities and challenges in our AI2050 motivating question,” said James Manyika, co-chair of AI2050 and a senior vice president at Google. “This technology can and will bring about an epochal shift in our society — and the AI2050 fellows are shaping that change so it is a benefit for all people.” 

is a non-profit organization founded in 2024 by Eric and Wendy Schmidt that works to accelerate scientific knowledge and breakthroughs with the most promising, advanced tools to support a thriving planet. The organization prioritizes research in areas poised for impact including AI and advanced computing, astrophysics, biosciences, climate, and space—as well as supporting researchers in a variety of disciplines through its science systems program. 

With files from Schmidt Sciences. 

Students from across the Faculty of Applied Science & Engineering and the Faculty of Arts & Science are acquiring industry-ready skills and making meaningful industry connections, thanks to a $1.38 million gift from the Royal Bank of Canada. 

The visionary support will enable students to delve deeper into topics the tech industry is confronting today with the Tech@RBC Insider Series, which features 12 learning sessions over the next three years. The gift will also create two powerful scholarships: the RBC Tech Scholars in AI Engineering and RBC Tech Scholars in Computer Science. Each valued at approximately $25,480, the awards will alleviate financial burden and transform the lives of 30 promising third-year undergraduate students over the next three years. 

“At RBC, we know students are critical to our future, forming the next generation of tech leaders and innovators,” says Martin Wildberger, executive vice-president of innovation & technology at RBC.  

“Our partnership with the University of Toronto is focused on helping motivate and encourage early talent to grow their skills beyond the classroom and learn from RBC’s technology leaders. Canada is home to some of the best and brightest students, and we aim to inspire and empower them to shape the future of technology for all of us.”  

Senior leaders at both faculties reflected on the significance of the gift and expressed gratitude. 

“It’s reassuring to know RBC shares our passion for ensuring brilliant students grow their skills to make an impact,” says Professor Deepa Kundur, chair of the Edward S. Rogers Sr. ������Ƶ of Electrical & Computer Engineering.  

“Thank you for your vision and dedication to empowering the next generation of tech talent right here at the University of Toronto.”  

“We are deeply grateful to RBC for this generous investment in our students and community,” says Professor Eyal de Lara, chair of the ������Ƶ of Computer Science.  

“By supporting the Tech@RBC Insider Series and new scholarships, this gift will open doors for our students to connect with leading voices in technology while reducing financial barriers to their education. It’s a powerful way to help our students thrive and contribute to the future of innovation.” 

In October, U of T Engineering and Arts & Science students packed the second-floor event space at the Schwartz Reisman Innovation Campus for the inaugural Tech@RBC Insider session, Cybersecurity: Defend the Digital Fortress. Milos Stojadinovic, senior director of advanced threat operations and distinguished engineer at RBC, kicked off the evening with a behind-the-scenes look at how banks tackle cybersecurity. Following a networking session, workshop participants rolled up their sleeves to tackle hands-on threat modelling and threat response simulation exercises. 

Many students, including Chloe Kentebe (Year 2 CompE), gained valuable insights from the session.

She was drawn to the lecture and workshop by her strong interest in cybersecurity mechanisms, and how they are designed and implemented in the financial space. Last summer, under the supervision of Kundur’s lab, she took on a research project aimed at understanding the cyber-physical security of autonomous vehicles.

She says this experience, as well as navigating her classes, participating in extracurriculars — including contributing to U of T Formula Racing as a deep learning team member — and attending events like the Tech@RBC session, have deepened her interest and broadened her understanding of cybersecurity and safety.

“To ensure the strength and resilience of a system, one needs to have a certain level of technical knowledge surrounding the dynamics of its environment, but it’s even more essential to have a mindset that can consider the unique complexities and edge cases related to the ways that the system can be infiltrated,” she says.

“The art and science of developing innovative and applicable solutions is a skill I commit to continuously improving through my education and extracurriculars.”

Meanwhile, fellow attendee Tuğra Canbaz felt a personal connection to the lecture and workshop.

from Türkiye and first-year student hoping to pursue a double major in computer science and economics, has seen the devastating effects of cybersecurity breaches in his home country.

“I can’t help but be interested in cyber security and regulations surrounding it,” says Canbaz, who is aiming for a career in tech, perhaps in financial technologies or cybersecurity.

“It’s also important to consider potential interactions with AI. Imagine if an AI algorithm was trained on leaked data and how invasive that would be. That’s something I want to work on safeguarding against in the future.”

The Tech@RBC hands-on lecture, workshop and networking session also put him in the proper frame of mind to consider future trajectories.

“I like solving problems creatively and I also like the social aspects of the job — working with people, putting humans at the centre of computer problem solving,” he says. “That’s what inspires me to do more.” 

 — Original story by Rebecca Cheung for

Interested in attending the next Tech@RBC Insider Session, co-hosted by Tech@RBC, the Faculty of Arts & Science and U of T Engineering? 

Mark your calendars and stay tuned for more information about sessions: 

: November 26, 2025 

Product ownership: February 3, 2026 

Technical careers: March 19, 2026 

Assistant Professor , who joined the ������Ƶ of Computer Science in January, has received funding and a for her work in privacy-preserving cryptographic systems.

Hithnawi leads the , where she develops technologies that protect sensitive data without sacrificing usability or performance. Her research explores ways to keep data private and secure while still allowing people and systems to use it effectively.

She designs compliers and frameworks that make advanced cryptographic techniques more efficient and accessible, demonstrating how rigorous privacy protections can be integrated into real-world systems.

The Google Research Scholar Program supports early-career faculty conducting research in areas relevant to Google. The Intel Outstanding Researcher Award recognizes academic researchers whose work significantly advances the future of technology.

Before joining the University of Toronto, Hithnawi was a faculty member at ETH Zürich, where she founded the PPS Lab. She has supervised graduate students, published extensively in top venues and helped shape the global conversation around secure, privacy-aware computation.

Associate Professor has been to the ’s (RSC) College of New Scholars, Artists and Scientists, a seven-year membership awarded to individuals who demonstrate exceptional achievement early in their careers.

Wang holds a joint appointment in the ������Ƶs of and Computer Science at the University of Toronto. He is also the chief artificial intelligence scientist at the University Health Network, a CIFAR AI chair at the and the senior vice president and head of biomedical AI at Xaira Therapeutics.

His research focuses on machine learning, computational biology and computer vision with an emphasis on applications in biomedicine.

“Bo’s interdisciplinary approach to artificial intelligence in medicine reflects the innovative nature of our department,” said Eyal de Lara, professor and chair of the ������Ƶ of Computer Science. “His contributions to AI in medicine are not only advancing the field but also improving patient care.”

, professor of chemistry and computer science at the University of Toronto, has been a Fellow of the (RSC), one of the country’s highest academic honours.

Aspuru-Guzik joined U of T in 2018, drawn by Canada’s commitment to innovation and social democracy. He said the recognition is especially meaningful as he prepares to become a Canadian citizen.

“I decided to move to Canada and continue my career at the University of Toronto, and it was a great decision,” he said. “I love Canada in many ways — from the university culture to the city and country culture. The fact that the Royal Society of Canada selected me as a fellow is very special.”

Aspuru-Guzik recently passed the Canadian citizenship test and is awaiting his citizenship ceremony. “It’s a great thing to have happen in 2025,” he added.

The RSC celebrates both academic excellence and societal impact. Aspuru-Guzik’s work bridges both, with research that advances scientific discovery and addresses global challenges.

As co-director of the with Varinia Bernales, Aspuru-Guzik leads efforts to accelerate discovery in chemistry and adjacent fields. His team contributed to the first AI-discovered lead candidate for a drug, published in , and helped pioneer organic flow batteries, featured in . Most recently, his lab demonstrated how AI and self-driving labs can accelerate molecular discovery, using organic lasers as a case study in .

“Our mission is to democratize self-driving laboratories via the ,” he said. “This work can help others develop materials for society and tackle challenges like climate change and technological equity.”

In 2023, the Acceleration Consortium received the largest federal research grant ever awarded to a Canadian university. The $200-million grant by the (CFREF) supports the consortium’s work on self-driving labs — combining AI, robotics and advanced computing.

“Alán’s work exemplifies the kind of excellence and innovation we value at U of T,” said Eyal de Lara, professor and chair of the ������Ƶ of Computer Science. “His contributions have had a profound impact on both academia and society. We’re proud to celebrate his recognition by the Royal Society of Canada.”

Aspuru-Guzik also mentors the next generation of researchers. His advice to early-career scientists: pursue bold, interdisciplinary questions.

“Follow what I call the Aspuru-Guzik/Whitesides rules,” he said, referring to a in . “Chase unsolved, yet achievable problems that could impact humanity. Don’t follow the crowd, focus on the unknown and work at multidisciplinary interfaces.”

Two University of Toronto computer scientists have received from the Government of Ontario for their innovative work in sustainable AI and next-generation imaging systems.

Associate Professor and Assistant Professor have each been awarded funding in the latest round of the competitive program.

Here is a closer look at the awarded projects:

Murat Erdogdu
Associate Professor, ������Ƶ of Computer Science and ������Ƶ of Statistical Sciences

Project Title: Training Neural Networks Efficiently with Scaling Laws and Simplicity Bias

It is imperative that we develop efficient methods to train large-scale AI systems that require less power and energy. This research aims to find patterns in data to optimize AI model training and ultimately reduce the computational cost and environmental impact of AI models. The project supports Canada’s growing investment in AI infrastructure by helping researchers build AI systems more efficiently.

David Lindell
Assistant Professor, ������Ƶ of Computer Science

Project Title: Single-Photon Imaging with Large Generative Models

Next-generation imaging systems are being developed to operate in extreme conditions such as near-total darkness or at very high speeds. This research combines ultra-sensitive cameras that detect single particles of light with powerful AI models that can generate realistic images. The resulting systems could improve technologies ranging from smartphone photography to autonomous vehicles and environmental monitoring, leading to safer, smarter and more capable imaging tools across Ontario.

Originally designed to render graphics for gamers and video editors, graphics processing units (GPUs) have evolved into the core computing engines that power today’s artificial intelligence (AI) models and cloud-based machine learning services.

Recognizing GPUs’ increasing importance, a team of computer scientists at the University of Toronto set out to test their vulnerability to a critical hardware attack known to affect the memory in central processing units (CPUs).

Their research shows for the first time that these Rowhammer-style attacks are effective against GPUs with GDDR memory, which is commonly found in graphics cards.

In turn, AI models that run on these GPUs are at risk of “catastrophic brain damage” — a degradation in accuracy from 80 to 0.1 per cent, according to , assistant professor in the ������Ƶ of Computer Science. That has significant implications for the accuracy of AI applications that rely on these models, like a hospital’s medical imaging analysis or a bank’s fraud detection system.

In a Rowhammer attack, memory cells are tricked into flipping bits — tiny pieces of data — by rapidly accessing nearby rows of cells over and over. This creates electrical interference that causes errors in parts of the memory the attacker didn’t directly touch, potentially allowing them to bypass security or take control of a system.

“Traditionally, security has been thought of at the software layer, but we’re increasingly seeing physical effects at the hardware layer that can be leveraged as vulnerabilities,” said Saileshwar.

Saileshwar, alongside second-year computer science PhD student Chris (Shaopeng) Lin and fourth-year computer science undergrad Joyce Qu, developed the proof-of-concept ‘’ attack on the GDDR6 memory in an NVIDIA RTX A6000, a GPU widely used for high-performance computing. Their has been accepted to USENIX Security 2025, a top-tier computer security conference.

They found that a single bit flip to change the exponent of an AI model’s weight could cause a massive reduction in the model’s accuracy.

“This introduces a new way AI models can fail — at the hardware level,” said Saileshwar.

The GPU users most at risk are those managing cloud computing environments, rather than individual home or office users. That’s because in the cloud, multiple users could be accessing a particular GPU at the same time, allowing an attacker to tamper with another user’s data processing.

The researchers’ attack had to account for the differences between CPU and GPU memory, Saileshwar explained. GPUs are tougher targets due to their faster memory refresh rates, slower memory latency and other architectural differences. Ultimately, the researchers leveraged the GPU’s parallelism, or its ability to run multiple operations simultaneously to optimize their hammering patterns. That adjustment led to the bit flips that demonstrated a successful attack.

But it wasn’t easy. “Hammering on GPUs is like hammering blind,” Saileshwar said, noting that they nearly gave up after failing to trigger any bit flips.

On CPUs, there are tools that help researchers inspect the memory interface, which help researchers understand how memory accesses behave and how instructions are sent from the CPU to memory. But because GPU memory chips are soldered onto the GPU board, there is no easy way to perform a similar inspection, he explained. The only signal they had was observing the bit flips that they eventually triggered.

The researchers privately disclosed their findings to NVIDIA earlier this year, and the company issued a to its customers in July.

NVIDIA’s suggested remedy, enabling a feature called Error Correction Code (ECC), can repel a GPUHammer attack, but at the cost of slowing down machine learning tasks by up to 10 per cent, the researchers found. And while the affected GPUs have some built-in defenses, GPUHammer was able to get past them, showing that the current generation of mitigations isn’t foolproof. Future attacks involving more bit flips might be able to overwhelm even the ECC’s remedies, they noted.

The findings highlight a need for greater attention to GPU security, work that is “just beginning,” said Saileshwar.

“More investigation will probably reveal more issues,” he noted. “And that’s important, because we’re running incredibly valuable workloads on GPUs. AI models are being used in real-world settings like health care, finance and cybersecurity. If there are vulnerabilities that allow attackers to tamper with those models at the hardware level, we need to find them before they’re exploited.”

Driven by a vision of ethical and dependable technology, is advancing AI systems that are as transparent as they are powerful, to safely serve people in daily life.

Bo, a second-year computer science PhD student in the , has been awarded a prestigious 2025 .

The scholarship provides $150,000 over three years to students who demonstrate leadership and a high standard of scholarly achievement in graduate studies at Canadian universities.

Bo's research focuses on improving how people understand and rely on AI, especially as generative tools like ChatGPT become more common. These systems can produce misleading or incorrect information, which may lead to over-reliance and poor decision-making. To address this, Bo is developing strategies to help users better understand and appropriately trust AI systems. Her work includes studying how people interact with AI and designing interventions to improve those interactions. She also explores ways to help users form accurate mental models of how AI works. She is supervised by Associate Professor .

“It's both exciting and challenging to contribute to the safe and responsible use of AI in everyday life,” says Bo.

“Receiving the Vanier is incredibly meaningful and knowing that my work is supported is deeply validating,” she adds. “I’m excited to focus my PhD research on improving how we design and evaluate AI tools — especially as they become integral to decision-making that impacts people.”

Weather prediction systems provide critical information about dangerous storms, deadly heatwaves and potential droughts, among other climate emergencies.  

But they’re not always accurate. And, ironically, the supercomputers that generate forecasts are also energy-intensive, contributing to greenhouse gas emissions while predicting increasingly erratic weather caused by climate change.  

“The process right now is very computationally expensive,” says James Requeima, a post-doctoral researcher in computer science at the University of Toronto and the .  

Enter Aardvark Weather, a weather prediction model developed by Requeima and other researchers using artificial intelligence (AI). Described in , the system produces results comparable to traditional methods, but is 10 times faster, uses a tiny fraction of the data and consumes 1,000 times less computing power.  

In fact, the model can be run on a regular computer or laptop. It’s also open-source and easily customizable, allowing small organizations, developing countries or people in remote regions to input the data they have and generate local forecasts on a minimal budget. 

The development could be a timely one. As ,  and , there’s a clear need for accessible and accurate weather forecasting around the world.

“You hear a lot about the promise of AI to help people and hopefully make humanity better,” Requeima says. “We’re hoping to enact some of that promise with these weather prediction models.” 

Aardvark Weather is being developed at Cambridge University — where Requeima completed his PhD in engineering and machine learning — and the Alan Turing Institute. Requeima joined the project in 2023. He received post-doctoral funding for the project last year from U of T’s , an .  

U of T News recently spoke to Requeima about the project and his role. 

How is weather currently predicted? 

The big weather forecasters, such as the  and the , take initial conditions representing the current state of the atmosphere and put that information into a supercomputer. They then run a numerical simulation and propagate that forward into the future to get forecasts of the future states of the atmosphere.  

Then they take observations from real-world sensing instruments and incorporate them into their current belief about the atmosphere and re-run the forecast. There’s a constant iterative loop. From these atmospheric predictions, you can build a tornado forecaster or a precipitation forecaster. 

How can AI do better and with less computing power? 

End-to-end deep learning fundamentally changes how we approach weather prediction. Rather than the traditional, iterative process that relies on expensive numerical simulations, we train our model to map directly from sensor inputs to the weather variables we care about. We feed in raw observational data — from satellites, ships and weather stations — and the model learns to predict precipitation, atmospheric pressure, and other conditions directly. While training the initial model requires computational resources, once trained, it’s remarkably efficient. The resulting system is lightweight enough to run on a laptop, making predictions orders of magnitude faster and more accessible than traditional supercomputer-based methods.

This means communities can deploy these models locally to generate their own forecasts for the specific weather patterns that matter to them.

Have others used AI for weather prediction? 

Machine learning has been applied to climate modelling before, but previous approaches still depended on numerical simulations as their input. Our key breakthrough is demonstrating that you can move out of this paradigm and map directly from observation to targets. This proof of concept opens up a fundamentally new approach to forecasting — we’ve demonstrated that accurate weather prediction doesn’t require supercomputer simulations as an intermediate step.

How can this technology be used in practice? 

We are open sourcing this model — making it available to the community so others will improve upon our model to make changes and train it to do local modelling. We’re hoping this will help democratize weather prediction.  

Forecasting quality is correlated with wealth, so developing nations don’t have access to as good forecasting as wealthier nations do. If we can help bring high-quality forecasting to areas that don’t have it before, that’s a really big positive of this work.  

, an associate professor of computer science in U of T’s Faculty of Arts & Science] — my adviser — and I want to use AI in positive ways. Climate prediction is an important tool for assessing and developing ways of dealing with climate change — and the better climate models we have, the better our science can be around tackling that problem. That’s a driving motivation for me. 

What was your contribution to this work? 

During my PhD, I worked on neural processes — a type of neural network model that is effective for numerical forecasting. We discovered it was well-suited for scientific applications, especially climate modelling. For Aardvark, I helped design the model architecture and the multi-stage training scheme. 

Where did the name Aardvark Weather come from?  

The first author on this research, Anna Allen from Cambridge, did a lot of the heavy lifting on this — which is going out and finding the data sources, including a lot of Canadian data from weather stations, weather balloons and ship observations. She’s from Australia and is a lover of interesting animals like sloths — and aardvarks.  

— Original story by Diane Peters for

Does artificial intelligence have subjective experience? Could AI outsmart and outmanoeuvre humans? What can Canada do to ensure it remains a leader in the global AI race that it helped kickstart?

These were some of the questions addressed by the University of Toronto’s Geoffrey Hinton — a emeritus of computer science and recipient of the — during a recent lecture and fireside chat held at Convocation Hall during the inaugural .

event saw the “godfather of AI” put forth two of his most compelling and controversial contentions: that large language models (LLMs) such as ChatGPT and others understand language — rather than merely regurgitate it — and that AI could pose an existential risk to humanity.

The lecture culminated in a lively exchange between Hinton and his former protégé Nick Frosst, a U of T alumnus and co-founder of AI language processing startup Cohere. The pair discussed and debated the promise and risks of the transformative technology.

On the subject of understanding, Hinton insisted LLMs can have subjective experience and are “quite close to [humans]” in terms of consciousness. Frosst, on the other hand, characterized such systems as “more conscious than a rock and less conscious than a tree.”

“It is very difficult to come on stage and disagree with a Nobel laureate,” he later joked, prompting laughter from the audience.

The conversation, moderated by CBC tech journalist Nora Young, was among the most anticipated at Toronto Tech Week, which ran from June 23-27 and from Toronto’s thriving AI and tech ecosystem.

“At the heart of that ecosystem sit our region’s excellent universities, with the University of Toronto the main catalyst,” said U of T President Meric Gertler in his introductory remarks, pointing to tech magazine highlighting tech founders and companies — a third of whom were connected to U of T.

President Gertler cited the — which Hinton co-founded — and the new as prime examples of U of T’s role as a key node in Toronto’s tech and innovation ecosystem. “The potential for discovery, invention and innovation at U of T and in the Toronto region is huge and inspiring,” he said.

The event also saw , with the financial institution set to continue supporting the Desjardins Speaker Series, the Desjardins Startup Prize — part of the annual U of T Entrepreneurship Week — and financial literacy workshops for a further three years.

During his lecture, Hinton traced the evolution of LLMs from his early experiments in the 1980s to today’s powerful systems. He then set out his argument that LLMs understand language and have subjective experiences, drawing on elements of philosophy, neuroscience and computer science to make his case.

He also reiterated his warning about the risks posed by AI, outlining two major concerns: the misuse of AI by bad actors, and the possibility of super-intelligent AI systems acting independently of human control.

Hinton later expressed concern about tech companies resisting AI regulations, comparing their stance to oil companies opposing environmental oversight. He noted that without adequate regulation, AI agents could cause problems in a number of ways. For example, he said, “they will be able to [invent] creative new ways of finding people’s passwords.”

He also suggested AI systems could have a major impact on the job market in the coming years.

Frosst, for his part, agreed on the importance of AI safety, but took issue with Hinton’s view on the extent and specific nature of the risks.

The conversation also touched on Canada’s place in the global AI revolution. Hinton praised recent initiatives such as the , but called for more proactive engagement across the business sector and government.

Frosst voiced optimism about Canada’s AI future.

“We invented this technology,” he said, noting Hinton’s foundational contributions to the field. “Canada has every right to be a leader in it.”

— Original story by Rahul Kalvapalle for