Friday, July 10, 2026

Meta bets $13 billion on Alberta with first Canadian data centre



Nate Glubish, Alberta’s minister of technology and innovation (left), and Alberta Premier Danielle Smith speak to press following the announcement on July 8, 2026. — Photo by Jennifer Friesen, Digital Journal

Meta is building its first Canadian data centre in Alberta, a campus worth more than $13 billion that the province calls one of the largest private investments in Canadian history and Meta’s biggest anywhere outside the United States.

Alberta Premier Danielle Smith announced the project at a news conference in Calgary on Wednesday, midway through the Stampede, alongside two ministers, the mayor of Sturgeon County and Meta’s data centre chief.

The 2.9-million-square-foot facility will sit on 1,750 acres in Sturgeon County, inside Alberta’s Industrial Heartland. It will be powered in part by the Greenlight Electricity Centre, the $4.6-billion natural gas plant that Pembina Pipeline, Morgan Stanley Infrastructure Partners and Kineticor announced last week.

The province says the campus will create more than 3,000 construction jobs and 300 permanent ones, and return about $250 million a year to Albertans through royalties, taxes, levies and fees. Meta is putting roughly $60 million into local roads and water infrastructure.

Alberta Minister of Technology and Innovation Nate Glubish, speaks at The Princeton in Calgary on July 8, 2026. — Photo by Jennifer Friesen, Digital Journal
Alberta Minister of Technology and Innovation Nate Glubish, speaks at The Princeton in Calgary on July 8, 2026. — Photo by Jennifer Friesen, Digital Journal

Alberta courted Meta for two years and closed the deal without incentives

The first phase draws 970 megawatts of electricity, enough to power a mid-sized city. The campus is designed to grow to 1,800 megawatts, more than the city of Edmonton uses at peak.

The co-located Greenlight plant is what lets the site reach that ceiling without drawing more from the public grid. Under Alberta’s rules, Meta pays for the new power generation and the grid upgrades its own connection requires, rather than passing those costs to other customers.

The province says that arrangement will lower bills for everyone else.

“In this case, in this project, Albertans could see a reduction of up to 6% on the transmission portion of their utility bill,” said RJ Sigurdson, Alberta’s minister of affordability and utilities.

RJ Sigurdson, Alberta minister of affordability and utilities. — Photo by Jennifer Friesen, Digital Journal
RJ Sigurdson, Alberta minister of affordability and utilities. — Photo by Jennifer Friesen, Digital Journal

Every Albertan who uses electricity pays a share of the cost of building and maintaining the province’s high-voltage grid, the network of transmission lines that moves power across long distances. On a power bill, that cost is the transmission charge and the total is roughly fixed, so it gets divided among everyone connected to the system. 

Sigurdson said that when a customer as large as Meta joins the grid and takes on a big portion of that cost, the amount left for households and businesses to cover shrinks, which is where the 6% reduction comes from.

“Without this grid connected pathway, Alberta may not have landed this massive new tenant,” he said.

Meta's Gary Demasi speaks to press in Calgary. — Photo by Jennifer Friesen, Digital Journal
Meta’s Gary Demasi speaks to press in Calgary. — Photo by Jennifer Friesen, Digital Journal

Nate Glubish, minister of technology and innovation, said he first met Meta’s data centre team two years ago at the company’s Silicon Valley headquarters, where Alberta was not yet on its list of places to build.

The province’s answer was a system it calls the concierge, a single provincial contact that walks a company through the electricity regulator, the grid operator, the utilities commission and the municipality at once instead of leaving it to file with each separately. Glubish said that team is now working with about 60 more projects, several of them at the gigawatt scale of the Meta campus.

Glubish said the province used no grants, tax credits or incentives to attract Meta, and that Alberta instead wrote a levy so data centres pay a share back to the public. He said Meta spent two years and hundreds of millions of dollars on engineering, permitting and consultation before the announcement.

Nate Glubish is Alberta's minister of technology and innovation. — Photo by Jennifer Friesen, Digital Journal
Nate Glubish is Alberta’s minister of technology and innovation. — Photo by Jennifer Friesen, Digital Journal

He also drew a line between Meta and projects that have failed in Alberta. Regulators rejected the Synapse proposal near Olds, and Rocky View County council turned down a separate campus after a hearing that ran most of a day.

“Meta is the customer, they are the user, they are the engineer, they are the designer, they are the builder, they are the operator,” Glubish said. “Whereas the proposed project in Olds is a developer who has an idea and believes that if they build it, someone will come and buy it.”

Smith repeated the pitch Alberta has been making for more than a year: affordable electricity, flexible power generation, a cool climate that eases the job of cooling servers, a skilled workforce and homegrown AI expertise.

“Artificial intelligence is transforming the global economy, and Alberta is making sure we lead rather than follow,” said Premier Danielle Smith. “We created the right conditions to attract world-leading investments while protecting the interests of Albertans.”

Alberta Premier Danielle Smith speaks at The Princeton in Calgary on July 8, 2026. — Photo by Jennifer Friesen, Digital Journal
Alberta Premier Danielle Smith speaks at The Princeton in Calgary on July 8, 2026. — Photo by Jennifer Friesen, Digital Journal

Meta’s pledges, and its record elsewhere

Water has been a flashpoint in Alberta’s data centre fights. The servers that fill these buildings run hot, and many centres cool them by evaporating large volumes of water, pulling from the same supply households and farms rely on.

Meta led with water for that reason.

The company said its Sturgeon campus, which does not have a completion date yet, will use a closed-loop liquid cooling system with dry cooling, which recirculates a fixed volume of liquid instead of constantly drawing new water, and needs no operational water for cooling. On-site water is limited to washrooms, fire protection and equipment maintenance.

“The water that’s used by this data centre, it’s less than a typical golf course in Alberta over the course of a year,” said Gary Demasi, vice-president of data centre strategy and development at Meta.

Gary Demasi is the vice-president of data centre strategy and development at Meta. — Photo by Jennifer Friesen, Digital Journal
Gary Demasi is the vice-president of data centre strategy and development at Meta. — Photo by Jennifer Friesen, Digital Journal

Meta committed to being water positive by 2030, meaning the company will fund restoration and conservation work that returns more water to local watersheds than the site consumes. Its first step in Alberta is a partnership with the conservation group ALUS and local farmers to protect 200 acres of grassland, trees and wetland in the North Saskatchewan River watershed, land that stores and filters water for the basin.

“We look forward to putting down roots in this community and building a strong and positive partnership for many years to come,” he said.

Meta has made versions of that promise where it has built before, and the record is mixed. 

In Richland Parish, Louisiana, the site of Meta’s Hyperion campus, the company secured the land through a shell company and pushed state tax legislation forward under non-disclosure agreements before the public knew it was involved, according to a Bloomberg investigation. After regulators approved the gas plants built to power the campus, consumer and environmental groups asked the state to investigate whether ratepayers would be left exposed if Meta walked away.

In Alberta, the rules came first. 

The province set its data centre requirements before opening the door to investment, making projects bring their own power and pay for their own infrastructure, and the county zoned and reviewed the Sturgeon site before the announcement, so Meta is funding its own power and grid work rather than leaving costs to other ratepayers.

The site also sits inside a zone set aside for heavy industry decades ago, with buffer zones between the plants and homes and no farmland or housing on the land itself. Sturgeon County Mayor, Alanna Hnatiw, said the county held the project to its controls on power, water and land use.

Meta said it will offset the campus’s electricity use by paying for an equal amount of clean power to be added to the grid elsewhere, a common industry practice, though the electricity that actually runs the site comes largely from natural gas.

Alanna Hnatiw, Mayor of Sturgeon County. — Photo by Jennifer Friesen, Digital Journal
Alanna Hnatiw, Mayor of Sturgeon County. — Photo by Jennifer Friesen, Digital Journal

The country’s biggest AI campus

Smith said the facility will power Meta’s own technologies, including Facebook, Messenger, Instagram, WhatsApp and its AI products. 

Last week Meta said it may start selling excess computing capacity to outside customers, a business it is still weighing. But no one at the announcement said whether Canadian companies would be able to buy time on the Sturgeon campus.

Data sovereignty is the deeper question, as putting the hardware on Alberta soil does not put it under Canadian control. 

The U.S. Cloud Act lets American authorities compel a U.S. company like Meta to hand over data it controls no matter where that data physically sits, so a data centre in Sturgeon County does not by itself keep what runs inside it under Canadian law.

The Government of Canada’s own cloud policy states that as long as a provider operating in Canada is subject to the laws of a foreign country, Canada will not have full sovereignty over its data. Storing data in Canada gives residency but sovereignty depends on who controls the data and which laws bind the company that holds it. 

At the press conference today, Glubish said Meta is the first of many projects the province expects, and it now has a reference customer to point to as it courts the rest.

Alberta puts $50 million behind Amii’s AI research


Cam Linke, Amii CEO, speaks at the funding announcement at Platform Calgary.
 — Photo by Jennifer Friesen, Digital Journal

Today, the province of Alberta announced a $50 million investment over five years for the Alberta Machine Intelligence Institute (Amii), home to two decades of reinforcement learning research.

Aimed at accelerating AI adoption across public services and industry, part of the plan includes leveraging the newly-established Alberta Intellectual Property Office to ensure IP is owned, protected, and commercialized in the province. 

Amii is one of Canada’s three national AI institutes, alongside Mila in Quebec and the Vector Institute in Ontario, operating under the Pan-Canadian AI Strategy. Established in 2002, it has a focus on reinforcement learning, a field in which the province is recognized as a world leader.

Alberta Premier Danielle Smith announced $50 million in funding for Amii. — Photo by Jennifer Friesen, Digital Journal

Five ministries, one investment

When a government funds something like this, it’s often one ministry writing the cheque. 

This time it’s five.

The ministries of Technology and Innovation and Advanced Education are co-leads, investing $15 million each, with Assisted Living and Social Services at $10 million. Primary and Preventative Health Services and Education and Childcare are each contributing $5 million.

“Alberta has been a leader in the AI space for decades, and the Alberta Machine Intelligence Institute… has played a major role in keeping Alberta on top,” said Alberta Premier Danielle Smith, announcing the funding at Platform Calgary. “They have worked closely with the government over the years to see how AI can be implemented into our everyday lives.”

She explained how the funding is the work of five ministries together, “because we know the significant impact that this could have on making the lives of Albertans better.”

Nate Glubish, Alberta’s minister of technology and innovation, credited Myles McDougall, Minister of Advanced Education as an equal partner on the file. 

Nate Glubish, Alberta’s Minister of Technology and Innovation, speaks at the Amii funding announcement at Platform Calgary. — Photo by Jennifer Friesen, Digital Journal

“I couldn’t have done it alone. I found some of the money in my budget, but my good friend and colleague [McDougall] here is an equal partner with me in that through his budget,” Glubish said. 

Glubish also pointed to the province’s history with Amii. “Alberta began investing in AI research more than two decades ago. That early bet paid off,” he said, noting how Amii’s chief scientific advisor, Dr. Richard Sutton, was a co-recipient of the 2024 A.M. Turing Award, frequently dubbed ‘the Nobel Prize of computing.’

“This $50-million investment is a vote of confidence in that team,” Glubish added. “It will put AI to work for the people of this province: better healthcare, faster public services, and more Alberta companies solving real problems.”

McDougall said that this technology is helping modernize systems for skilled tradespeople. 

“We apply AI to apprenticeship delivery, streamlining the path to graduation, cutting the paperwork for employers, and getting Alberta’s skilled trade system ready for the workforce demand that’s coming.”

He also gave credit to Alberta’s post-secondary system.

“Our universities, colleges, and polytechnics are training the next generation, the engineers, the entrepreneurs, researchers, and skilled professionals who will lead in an AI-driven economy,” he added. “This investment backs them and strengthens an ecosystem where students can learn, researchers can lead, companies can scale right here in Alberta.” 

Myles McDougall, Alberta’s Minister of Advanced Education, speaks at the Amii funding announcement at Platform Calgary. — Photo by Jennifer Friesen, Digital Journal

Bigger than a funding announcement

Evan Solomon, Canada’s Minister of AI and Digital Innovation, was in attendance at the funding announcement, using the stage to put Alberta’s investment in a national context. 

“We have classically, as a country, been great at research, and we’ve sucked at commercialization,” he said. “Alberta has led the way to show how we have to do that better.”

Solomon cited a commercialization gap, where companies often end up relocating outside of the country, taking their talent and patents with them. 

“We are losing our best IP and brains, so we are stopping that,” he added, later tying that loss to a broader argument about sovereignty. 

“If we don’t build the infrastructure here, we’re going to have to rent it from someone else,” he said. “If we don’t build the innovation here, we’re going to have to buy it from someone else. And if we don’t make the rules here, we have to follow someone else’s.”

This funding comes the day after the news that Meta is building its first Canadian data centre in Alberta.

It also comes on the heels of a Government of Canada and the Canadian Institute for Advanced Research (CIFAR) announcement made at the Upper Bound AI Conference in Edmonton in May. Solomon announced the appointment and renewal of 42 Canada CIFAR AI Chairs, backed by a $24 million investment, all affiliated with the three national AI institutes.

Evan Solomon, Canada’s Minister of AI and Digital Innovation, speaks at the Amii funding announcement at Platform Calgary. — Photo by Jennifer Friesen, Digital Journal

Canada’s recent “AI for All” national strategy goes further, pledging to strengthen the national AI institute network and increase the Chairs program from 130 to nearly 200 researchers. He outlined an investment of $130 million into the national AI institutes, and mentioned $700 million in affordable sovereign compute for Canadian SMEs, an expansion of the Compute Access Fund.

Finally, Solomon revealed a $1.8 billion talent attraction strategy to attract 1000 of the top researchers and their labs to Canada.

“We need the talent to come to Canada,” he said. “It’s the best place to be, it’s the best place to build, and it’s the best place to start a company.”

Building on two decades of research

According to the province, Alberta’s government has invested about $100 million in Amii since 2002. Through its work under the province’s first Technology and Innovation Strategy, Amii has worked with 786 companies to adopt or advance AI, supported 318 startups, and reached 1.2 million people through its literacy programs.

The institute currently has more than 500 active researchers and 71 Amii Fellows/Canada CIFAR AI Chairs.

“We and our strategy have promised to do a free AI literacy course for every Canadian, because we believe everybody should have access to literacy, and we’re going to give an AI tool and agent to every post-secondary student in Canada for five years, free and available, because they need tools to build,” said Solomon, giving kudos to Amii CEO Cam Linke on his work with training and education.

“What’s happened here in Alberta at Amii is the model for the nation.”

The new Amii funding is meant to work against the ‘brain drain” pattern Solomon described. With the Alberta Intellectual Property Office, patents developed with public funds are kept in the province, framing the investment as a way to grow new jobs, attract private investment, and keep ideas developed in Alberta from moving elsewhere.

Photo by Jennifer Friesen, Digital Journal

Linke said the investment builds on that foundation. 

“This investment enables Amii to strengthen the foundations of Alberta’s AI ecosystem by supporting startups, accelerating AI adoption across industry and the public sector, and building the AI literacy needed to prepare the workforce of the future,” he said. 

“Together, we are ensuring Alberta remains a destination for AI talent, innovation and investment while creating lasting economic and societal benefits for Albertans.”

Healthcare is a priority

Justin Wright, minister of primary and preventative health services, said the investment is meant to support faster, better-targeted care. 

“Alberta’s primary and preventative health system depends on getting the right information to the right provider at the right time,” he said. 

“By investing in Amii, we are supporting made-in-Alberta AI solutions that can help improve screening, earlier detection, system navigation and front-line decision-making. Used responsibly, this technology can reduce administrative burden, strengthen prevention-focused care and help Albertans access timely, high-quality services closer to home.”

Linke closed the announcement by tying the funding to Amii’s origins. 

“When you look at the history of Amii, it was an ambitious thing back in 2002 for the province to invest in AI,” he said. “We’re excited for what we can do with the province,  and what we’re going to do with this announcement, not just to be the place that powers AI globally, but also the place that produces the researchers, the talent, and the startups that are going to solve the world’s biggest problems right here.”

 

Humanitarians look to put the AI in aid

AFP
July 9, 2026 

The World Food Programme is using autonomous trucks to deliver aid in South Sudan, Sudan and Uganda – Copyright AFP JOEL SAGET

From remote-controlled trucks delivering life-saving aid in dangerous settings to mobile phone data analysis flagging mass displacement, humanitarians are eyeing ways in which artificial intelligence can speed up and improve their operations.

There have been plenty of warnings about the dangers of AI for aid agencies, who face growing challenges of securing often extremely sensitive data and swelling misinformation about their operations and beneficiaries.

But at the AI for Good summit in Geneva this week, a handful of humanitarian-focused displays emphasised the technology’s positive potential.

Parked in one corner of a vast hall at the Palexpo conference centre was a giant white SHERP vehicle, resembling a hulking Martian rover, decked out with cameras and sensors and a drone landing-pad on the roof.

Made in Ukraine, SHERPs are amphibious vehicles that can float on water, drive through swamps and flooded rivers with their giant wheels, and climb over obstacles up to one metre (3.3 feet) high.

The UN’s World Food Programme is preparing to begin field-testing a version of the AI-enabled truck that can be steered remotely through the most dangerous and difficult terrain to reach people in need.

“I think this could be a game-changer,” Bernhard Kowatsch, head of WFP’s global accelerator and ventures innovation division, told AFP.

The technology, he said, “should allow us essentially to reach people that otherwise never would have been reachable”.



– Not possible without AI –



WFP already has drivers using SHERPs to deliver aid in Sudan, South Sudan and Uganda.

But after numerous heartbreaking losses of drivers, it tasked the German Aerospace Center (DLR) to help equip the vehicles with AI and other technologies, making it possible to control them remotely through particularly dangerous terrain.

The idea is to set up a shipping container control room in a safe area, where a human can remotely control the vehicle on the last, most treacherous leg of its journey.

Tests have been conducted in Germany, and will be tried out in the field in Uganda in 2028, said Armin Wedler, who is coordinating DLR’s Autonomous Humanitarian Emergency Aid Devices (AHEAD) project.

Standing next to the 2.8-metre high vehicle, he told AFP that the team had used “remote-control technologies which are based on mathematics and old-school… research”, but stressed: “We would not be able to process everything without using also AI”.

It would be possible to make the vehicle fully autonomous, Wedler said, but stressed that in complex humanitarian settings “we have to have a human in the loop”.

“We’re not talking about driving on clear streets with clear lanes. There are no streets,” he said, also describing scenes where aid trucks are suddenly swarmed by desperately hungry people.

“There’s no AI autonomous algorithms ever capable to handle that safely.”



– ‘Life-saving’ –



Among more than 200 exhibitors at the summit — showing off everything from humanoid robots to bionic prosthetics and emotional companions — the other humanitarian displays were more discreet, with pamphlets detailing how AI tools are boosting and streamlining operations.

Among them, the UN refugee agency detailed a new Legal Virtual AI Assistant for lawyers and legal officers representing refugees, enabling them to swiftly determine the rights available within country-specific legal frameworks.

Rebeca Moreno Jimenez, the lead data scientist at UNHCR’s Innovation Service, told AFP that building cases faster and more efficiently can be “life-saving for many refugees”.

Another UN initiative called Data Insights for Social and Humanitarian Action, or DISHA, relies on partnerships with private actors such as Google and McKinsey to provide humanitarian organisations with data and AI models to speed up and improve disaster responses.

One project uses AI analysis of anonymised mobile phone data to spot mass-population movements during disasters, determining where people are fleeing, to help humanitarians better tailor their response.

Another uses AI for rapid analysis of satellite images taken before and after disasters like last month’s earthquakes in Venezuela to determine building damage.

The aim is to give humanitarians “accurate information early enough to make better decisions (and) avoid going to the wrong place when there are people who need you somewhere else”, DISHA product lead Andreas Kortis told AFP.

Global AI industry falls short on safety, think tank warns

AFP
July 8, 2026

Mistral ranked last in a survey on the management of risks associated with artificial intelligence – Copyright AFP Paige Taylor White

US artificial intelligence lab Anthropic scored the highest in a semiannual safety ranking, but globally the industry fails to combat “existential” threats, according to a report released on Tuesday.

Meta moved up two spots to fourth place, while xAI dropped three spots to seventh place, just ahead of China’s DeepSeek and France’s Mistral, which placed last, according to US-based AI safety think tank Future of Life Institute, which ranked nine of the world’s leading AI companies.

Seven researchers and governance experts determined the rankings based on public data and information provided by the companies.

They evaluated efforts across six distinct categories: risk assessment, current harms, safety frameworks, existential safety, governance and accountability, and information sharing.

No company received an “A” in any single category, while Anthropic got the best overall score of “C+.”

Mistral was included on the list for the first time, though when asked by AFP to comment on its last place, the company said the report’s framework isn’t suited for its approach to developing AI models.

The French company develops so-called open models, which allow users to download and modify them. Many of its competitors develop closed AI models — including Anthropic, OpenAI and Google DeepMind, which are also included in the report.

“I was very disappointed to find that they came last, especially since Europe has really…been a leader in AI safety,” Max Tegmark, an MIT professor and Future of Life president, told AFP.

“We reached out many, many times” but Mistral did not respond to the organization’s survey, Tegmark continued.

Alibaba, xAI and DeepSeek did not respond to its survey either, the organization said.

Three Chinese developers included in the report also produce open models and landed in the bottom half of the ranking: DeepSeek (fifth), Alibaba Cloud (sixth) and Z.ai (eighth).



– ‘Questionable’ practices –



The report noted that several companies that previously banned their technology from military uses have “gradually reversed course,” including Anthropic, which the report criticized for having “questionable military engagements.”

The US government used Anthropic’s technology in military operations in Venezuela and Iran over the past year, according to various media reports — though the company was subject to a recent ban by the Pentagon over disagreements on AI safety.

All nine companies are failing when it comes to combating “existential” threats such as pursuing models that reach human-level intelligence, known as “artificial general intelligence” or AGI, the report said.

Although “constructive attempts exist,” efforts across the board are “entirely inadequate.”

Other risks include the possible misuse of a model to carry out a cyberattack or perform tasks potentially harmful to humans.

Anthropic was thrust into the spotlight recently after it released its most powerful model yet, called Mythos.

In early April, the San Francisco-based company released Mythos only to a handful of trusted organizations due to its abilities to expose cyber safety vulnerabilities to bad actors.

However, by June 12 the US government blocked Anthropic from releasing Mythos to foreigners on national security grounds.

The Trump administration eventually lifted the ban a couple of weeks later on June 30.


To defend your software, first teach AI to break it



From the lab to the field and back, researchers and a former doctoral student develop artificial intelligence-driven tools to expose software vulnerabilities.




Virginia Tech





When Ying Zhang '23 was a doctoral student at Virginia Tech, she spent years learning to think like an attacker — probing software for the hidden weaknesses that developers miss and malicious actors exploit.

Now an assistant professor at Wake Forest University, Zhang has come back to that work and to the people who shaped it. She and her former doctoral advisors in the Department of Computer Science, Associate Professor Na Meng and Professor Danfeng "Daphne" Yao, have continued publishing influential research together, this time with a new cohort of Ph.D. students working alongside them.

The result is new research that posits that the best way for the cybersecurity community to guard against software vulnerabilities is by teaching artificial intelligence (AI) to identify and attack them. Their study about their approach will be presented on July 7 in Montreal, Canada, at the ACM International Conference on the Foundations of Software Engineering, one of the field's top-tier venues.

Finding the exploit in the haystack

At the heart of their work is a problem most people never see but that affects everyone: software vulnerabilities. Every time a person uses an app, buys something online, or interacts with a digital service, their data travels through invisible connectors called application programming interfaces, or APIs.

"You can view an API as a communication channel between two pieces of software," Zhang said. "If the API accepts malicious or unexpected inputs without performing proper validation and security checks, attackers can exploit those weaknesses to trigger vulnerabilities, compromise systems, or carry out successful cyberattacks."

These weak points are everywhere. Modern applications are built in layers, stacking third-party libraries and external code on top of one another in ways that make it nearly impossible for a developer to track every potential risk. A flaw buried deep in one layer can ripple upward, quietly exposing systems that developers didn't even know were at risk.

The problem isn't just technical. It's human. Developers, under constant pressure to ship working software, tend to treat security as an afterthought.

"Most of the time, security is considered a second-class citizen," Meng said. "Developers often prioritize functionality over security, especially when the risks are not immediately visible."

Raising awareness to protect systems

Meng and Yao's approach to closing that gap is as clever as it is counterintuitive: to help defend software, they teach artificial intelligence to attack it.

Working with Zhang and the current Ph.D. students in computer science, the team developed a system that uses large language models — the same underlying technology that powers tools like ChatGPT — to automatically generate what researchers call “proof-of-concept exploits.” These are step-by-step demonstrations showing exactly how a real attacker could take advantage of a known flaw.

The logic is simple but powerful. When a developer sees an abstract warning that their software has a vulnerability, they may ignore it or put off fixing it.

"If developers need a strong motivation, they say, 'Show me the exploit,'" Zhang said. "Then they will make the changes."

In testing, the system successfully generated proof-of-concept exploits with a high level of reliability. That’s a meaningful leap forward in solving a problem that has long resisted automated solutions. The goal, Zhang said, is entirely defensive: to help developers understand and address risks before malicious actors find them first.

The novelty of the approach has gotten some high-level attention. OpenAI has showed interest in the project, an early sign that it is resonating beyond academic circles.

Automating security

A second major thread of the team’s research targets the software supply chain. This complex web of dependencies connects modern applications. The team develops automated tools to identify which specific APIs within a piece of software are vulnerable, information that is often missing or incomplete and leaves developers without a clear picture of where to focus their security efforts.

"It's hard for developers to conduct code inspections or vulnerability localization, missing this information," Zhang said. "So we try to automatically derive it."

As software grows more interconnected, that precision becomes more important. Knowing a vulnerability exists somewhere in a system is far less useful than knowing exactly where it is and what it would take to exploit it.

Full circle

For the doctoral students on the team, working alongside Zhang carries a particular resonance. She was once exactly where they are now, learning the craft of cybersecurity research in the same labs, under the same advisors, asking the same kinds of questions.

“Working with Professor Zhang has taught me that good software engineering and security research takes patience and care,” computer science doctoral student and co-author Zhengjie Ji said. “She helped me see that results need to be solid enough for us to stand behind, and that writing is where we make the logic of the work clear to others.”

Wenjia Song '24 also worked on the project while a doctoral student. She went on to work at Google, Meng said.

For Meng and Yao, the continued collaboration with Zhang is a reminder of how sustained mentorship produces colleagues who return to push the work further.

"Dr. Zhang is a whisperer for software developers," Yao said. “She cares deeply about their day-to-day challenges. Her focus on creating innovative solutions for real-world needs is inspiring."

This work has been supported by the National Science Foundation, Office of Naval Research, and the Commonwealth Cyber Initiative. Yao is affiliated faculty in the Sanghani Center for Artificial Intelligence and Data Analytics.


Why governance lets Thomson Reuters

move faster on AI

David Potter
July 2, 2026



Thomson Reuters head of data and analytics Caitlin Halferty joins Snowflake EVP of product Christian Kleinerman on stage at Snowflake Summit. Photo courtesy of Snowflake

There’s pressure to get AI working fast, and governance often gets left for later. Build the pilot, prove it works, sort out who can see what data afterward.

Caitlin Halferty has seen where that leads. Teams get the pilot running, and then a problem forces the governance work they skipped, like finding people can see data they shouldn’t.

“They’ll build a really great data capability, perhaps a pilot, and then they’ll go to bolt on the governance later,” says Halferty, head of data and analytics at Thomson Reuters, in an interview with Digital Journal.

Thomson Reuters is a Toronto-based company whose customers are lawyers, accountants, and tax and audit professionals, the kind of work where being wrong carries a cost. Halferty’s argument is to govern the data first and then build AI on top of it.

“Governance for us has enabled us to de-risk and accelerate our AI transformation,” she said on stage at a Snowflake conference in June.


Getting every system to agree on what words mean


Good data isn’t enough on its own.


A company can connect every source and lock down who sees what, and its systems can still disagree on something as basic as how many customers it has.

“Unless you also invest in a semantic capability that really makes sense of that data, you still find yourself where you’ll arrive at conflicting answers on number of customers or best performing products in market,” says Halferty.

A semantic layer acts as a shared business dictionary, giving every system the same definition of terms like “customer” or “best-selling product,” so reports and AI tools don’t arrive at different answers.

In Thomson Reuters’ case, Halferty puts the system underneath at roughly 37,000 governed tables and 350 data sources feeding one secured source of truth.

The work began with finance and the team expanded from there. The reason? If finance will file its external reports off a data system, it vouches for the system everywhere else.

She said more than 1,500 people across finance and the business use that semantic layer for everyday decisions.

Work that took months now runs near real time, and a finance task that once meant waiting on Excel files that took 90 minutes to load is gone.

The agreed-upon data also enables AI agents to move faster. When every source already means the same thing by “customer,” an agent doesn’t stop to sort out the difference.

“Because we focused on it early, we’re able to move faster,” says Halferty.

Curated, governed data, she adds, is “sort of a gold mine for AI agents.”

Her test for whether the work is real is whether it has reached production.

“This is finance-validated metrics embedded in our key workflows,” she said.
A standard set by what the customer can’t get wrong

The standard for the data and the AI built on it is set by the people who buy from Thomson Reuters. Its customers are lawyers, accountants, and tax and audit professionals.

“Their name and reputation is on the line, it can’t be wrong,” says Halferty.

She calls the bar “fiduciary standard”. It means more than 175 years of curated content, thousands of subject-matter experts checking outputs, citations built into the product so users see where an answer came from, and a promise not to train models on customer data.

It also means testing before anything ships.

Halferty says every AI capability runs through the company’s responsible-AI practice, checked for hallucination and bias, with the security team testing for prompt injection, before a customer ever uses it.

Where to start without the scale

Thomson Reuters has scale, budget, and a long head start on clean data. Canada is pushing businesses to adopt AI, and boards are increasingly asking where to begin. For smaller companies, where’s the right place to start?

Halferty suggests starting by learning what your peers are doing first. Bring in outside help if the skill isn’t on staff. Then find where your own data lives.Thomson Reuters had customer data alone spread across 23 sources before pulling it into a single customer master.

“Figuring out how to pull that together across those different data sources, and leveraging a data platform and consulting capability, to identify what are you going after as an organization, what are therefore the data domains I need to prioritize,” says Halferty.

Get the data that matters most governed first, and the speed comes from there.
Final shotsCheck whether your systems agree on basic definitions before adding more data or AI on top. If two reports count customers differently, that gets worse with AI, not better.
Build access rules and governance into a project from the start. Adding them after a working pilot is where teams find people can see data they shouldn’t.
Find where your data actually lives before anything else. Halferty’s was spread across 23 systems, and pulling it into one place came before the AI work.


Revolutionizing material synthesis with artificial intelligence



Collaboration between the startup alqem AI and MPI CPfS



Max Planck Institute for Chemical Physics of Solids

Revolutionizing Material Synthesis with Artificial Intelligence 

image: 

Hanh Nguyen (CEO, left) and Julia Krez (VP Operations, right) examine an arc furnace. Julia Krez, a chemist, earned her Ph.D. under Prof. Claudia Felser and is also a member of the Board of Trustees of the MPI for Chemical Physics of Solids.  

view more 

Credit: © alqem AI GmbH





Collaboration with the Max Planck Society: The Max Planck Society and its institutes have long-standing research partnerships in the field of quantum materials and engage in active collaboration with industry and startups.

Innovative AI platform: Alqem AI's platform combines extensive databases of material predictions with high-quality training data and laboratory synthesis to turn digital predictions into real materials.

Partnership with Alqem AI: Alqem AI is launching a project to identify new magnetic materials that do not contain rare earth elements. The company is receiving support for this subproject from Prof. Claudia Felser, director at the Max Planck Institute for Chemical Physics of Solids.

 

Bronze, iron, steel, silicon: Every major technological era has been shaped by new materials. However, only a fraction of the theoretically possible crystalline compounds have ever been synthesized. Hundreds of millions of possibilities thus remain undiscovered. At the same time, the supply of critical raw materials is increasingly falling into the hands of just a few countries. High-performance permanent magnets – which are indispensable for electric vehicles, wind turbines, robotics, and defense systems, among other applications – are produced in China at a rate of about 90 percent. Recent export restrictions have made the supply of materials a key geopolitical issue.

The MPG and its institutes, which are active in the fields of solid-state research and related areas, have long been addressing this key issue and conducting research on so-called quantum materials. There are numerous scientific collaborations and agreements on these topics between the Max Planck Society and other scientific partners. The MPG has also long pursued scientific exchange between Max Planck Institutes and industry or startups. The most recent example is the scientific support provided to Alqem AI, a DEEPTech startup, by the Max Planck Institute for Chemical Physics of Solids in Dresden in the field of next-generation magnetic materials, led by Claudia Felser, director of the institute and vice president of the Max Planck Society.

Alqem AI, which has just completed a pre-seed round of financing totaling 8 million euros, is pursuing an approach that uses AI to identify suitable materials. The company was founded by the team behind Alexandria, the world’s leading open materials database, which is used by numerous universities and companies in AI-driven materials research. At the core of the AI platform are two proprietary data foundations: a database of predicted materials on an unprecedented scale, as well as high-quality training data for material properties that did not previously exist in this form. In-house laboratory capabilities for synthesis ensure that digital predictions are turned into real materials.

“We’re starting where the need is greatest: with rare-earth-free magnets – a product the world urgently needs, since rare earths have remained irreplaceable for decades. But the AI platform we’re building isn’t limited to a single class of materials,” says Hanh Nguyen, CEO of Alqem AI.

Claudia Felser’s team has been working in the field of synthesizing and characterizing rare-earth-free magnets for quite some time. Claudia Felser is committed to this startup out of conviction: “Discovering a truly new permanent magnet is one of the most difficult problems in materials science. The last real breakthrough was over forty years ago. ”What makes alqem’s approach special is that it doesn’t stop at prediction. In our collaboration, we’re putting precisely those rare-earth-free candidates through experimental testing that researchers have long been waiting for. And it’s the combination of large-scale computer-aided screening and systematic synthesis that convinces me this can work.”

***************************************************

The research at the Max Planck Institute for Chemical Physics of Solids (MPI CPfS) in Dresden, Germany, aims to discover and understand new materials with unusual properties.

In close cooperation, chemists and physicists (including chemists working on synthesis, experimentalists and theoreticians) use the most modern tools and methods to examine how the chemical composition and arrangement of atoms, as well as external forces, affect the magnetic, electronic and chemical properties of the compounds. New quantum materials, physical phenomena and materials for energy conversion are the result of this interdisciplinary collaboration.

The MPI CPfS (www.cpfs.mpg.de) is part of the Max Planck Society and was founded in 1995 in Dresden. It is divided into four departments and four independent research groups. The Institute employs around 280 staff, of whom about 180 are scientists including 70 doctoral students from more than 20 nations.

 

San Andreas Fault: Hidden movements revealed by artificial intelligence


New study uncovers previously undetected slow slip events beneath Parkfield in Central California and reveals their influence on seismic activity along the fault there




GFZ Helmholtz-Zentrum für Geoforschung

Deep-learning workflow for detecting slow slip events. 

image: 

Deep-learning workflow for detecting slow slip events. Continuous daily borehole strain observations (white curves) and their corresponding wavelet representations (background colors) from corrected strain components were used as input to the neural network. The red dashed box highlights a short-duration slow slip event recorded on 15 June 2013. The autoencoder with skip connections learns a compact latent representation of the input data, which is subsequently used for unsupervised clustering to distinguish deformation signals from non-deformation signals.

view more 

Credit: Zahra Zali, GFZ





When people think about geological faults, they usually think about earthquakes. Yet faults do not only move during earthquakes. Sometimes they slip silently, without generating noticeable shaking, releasing stress over hours or days through slow fault movements that remain largely hidden from conventional monitoring systems. Scientists have long suspected that these silent motions are an important part of the earthquake cycle. However, because they produce only subtle signals, they are notoriously difficult to detect. As a result, many questions remain unanswered: How often do they occur? Where do they happen? And can they influence subsequent seismic activity?

A research team led by Dr Zahra Zali (GFZ Helmholtz Centre for Geosciences), together with Prof. Patricia Martínez-Garzón (GFZ), Dr David Mencin (EarthScope), and Prof. Gregory C. Beroza (Stanford University), has uncovered a previously hidden population of so-called slow slip events beneath the Parkfield section of California’s San Andreas Fault. Using artificial intelligence and highly sensitive strainmeter observations, the researchers identified dozens of short-duration slow slip events and showed that these silent fault movements are systematically followed by increased low-frequency earthquake activity. The study has been published in Nature Communications.

Searching for hidden fault activity

Parkfield occupies a special place in earthquake science. Located on the San Andreas Fault, it is one of the most intensively monitored fault zones in the world. For decades, researchers have used Parkfield as a natural laboratory to investigate how faults accumulate and release stress. Despite the extensive monitoring network, some fault processes have remained remarkably difficult to observe.

“Faults can move in ways that do not generate strong seismic waves and therefore escape traditional earthquake detection methods,” says lead author Dr Zahra Zali. “We wanted to know whether important fault slip processes might be hidden within years of continuous deformation measurements.”

To search for such hidden signals, the team analysed continuous observations from borehole strainmeters. These instruments are capable of detecting extremely slow and small deformations in the Earth’s crust and are among the most sensitive tools available for monitoring active faults. The challenge, however, is that strainmeters produce enormous volumes of continuous data. Subtle transient signals can easily remain unnoticed among long-term deformation trends, environmental influences, and instrumental noise.

Artificial intelligence discovers what traditional methods missed

To address this challenge, the researchers developed a deep-learning approach capable of automatically identifying characteristic patterns associated with slow fault slip. Rather than searching for predefined signals, the artificial intelligence system learned directly from the continuous strain data and grouped similar deformation patterns together. This allowed the researchers to detect previously unrecognized short-duration slow slip events that release stress in a few hours.

“These events are difficult to identify by conventional methods because they are small and often hidden within complex background signals,” explains Zali. “Artificial intelligence allowed us to recognize their patterns that would otherwise have gone unnoticed.”

The analysis resulted in the first catalogue of short-duration slow slip events at Parkfield derived directly from continuous strainmeter observations. Independent observations from nearby creepmeters further supported the existence of these events. By estimating the location and direction of slip, the researchers found that the events occurred at shallow depth and were consistent with the right-lateral motion of the San Andreas Fault.

Silent slip and seismicity are connected

The discovery became even more intriguing when the researchers compared the timing of the newly detected slow slip events with low-frequency earthquakes (LFEs), a special class of weak seismic signals associated with fault slip processes. They found that low-frequency earthquake activity increases following the occurrence of slow slip events.

This observation suggests that even small episodes of aseismic fault motion can modify local stress conditions and influence subsequent seismic activity. “Our results show that these slow fault movements are not isolated phenomena,” says Prof. Patricia Martínez-Garzón, Working Group Leader in GFZ-Section 4.2 “Geomechanics and Scientific Drilling” and professor at RWTH Aachen University, who supervised the project. “They appear to be linked to changes in seismic activity, which suggests that slow slip may play an important role in how stress evolves along active faults.”

Filling a missing gap in the earthquake Science

Slow slip events have been studied extensively in subduction zones, where one tectonic plate dives beneath another. However, comparable observations in transform fault systems such as the San Andreas Fault have remained limited, particularly for short-duration events. The new study helps fill this observational gap in slow earthquake research. The researchers also found that the detected events follow the same relationship between event size (seismic moment) and event duration as observed for regular earthquakes. In other words, the way the size of these slow slip events scales with how long they last is similar to that of earthquakes. Together, these findings support the growing view that fault slip occurs across a continuum of behaviors ranging from silent deformation to destructive earthquakes.

A new window into hidden fault processes

The findings highlight the increasing role of Artificial Intelligence in Earth science and demonstrate how machine-learning approaches can reveal previously hidden signals within large geophysical datasets.

The researchers expect that similar short-duration slow slip events may exist on other faults worldwide and that future studies using dense geodetic monitoring networks could uncover additional examples.

“Many important fault processes occur without producing damaging earthquakes,” says Zali. “By detecting these hidden signals, we can gain a more complete picture of how faults behave between earthquakes and how stress is transferred through the Earth’s crust.”

 

Original study:

Zali, Z., Martínez-Garzón, P., Mencin, D. et al.: „Slow slip modulates low-frequency seismicity on the Parkfield segment of the San Andreas Fault.“ Nat Commun 17, 5137 (2026). https://doi.org/10.1038/s41467-026-74095-9

Chip titan SK hynix to set price for mega US listing

AFP
July 8, 2026
SK hynix has seen its profits and share price skyrocket – Copyright AFP Jung Yeon-je

South Korean chip giant SK hynix is expected on Thursday to set the price for its mega US listing that could be one of the world’s largest stock sales.

As governments and tech firms race to build data centres to train and run artificial intelligence, SK hynix — a supplier to industry behemoth Nvidia — has seen profits and its share price skyrocket.

The Asian semiconductor giant plans to issue the equivalent of nearly 18 million new shares on Wall Street’s tech-heavy Nasdaq index on Friday.

SK hynix said this week it hoped to raise a whopping 43 trillion won ($28 billion) by selling American Depositary Receipts (ADRs), which allow shares of foreign companies to be traded on US exchanges.

That goal was revised down from an initial target of more than 45 trillion won, but recent volatility in the company’s shares make firm estimations difficult.

Global tech stocks have tumbled in recent weeks amid fears about extended valuations — SK hynix has jumped more than 200 percent this year — and questions over when the enormous spending on AI will reap returns.

But the US listing has enjoyed considerable interest, with Bloomberg reporting it was more than seven times oversubscribed — and could be among the biggest ever.

The amount raised will likely not come close to the record $75 billion raised in SpaceX’s IPO last month, which made founder Elon Musk the world’s first trillionaire.

But it could rival Saudi Aramco’s 2019 $25.6 billion debut in the Gulf, and the $21.8 billion raised by Chinese tech firm Alibaba in its New York IPO.

Dilin Wu, research strategist at Pepperstone, told AFP the listing was “a huge development that should broaden the capital base for the memory sector”.



– High bandwidth –



SK hynix is already traded on Seoul’s Kospi index, where its market capitalisation soared past $1 trillion in May.

The milestone was also recently hit by domestic rival Samsung Electronics and US chipmaker Micron — with the AI boom thrusting the three semiconductor firms into a previously exclusive club of around a dozen companies, nearly all American.

Samsung, SK hynix and Micron dominate the global market for the advanced components known as high-bandwidth memory (HBM), used in AI servers alongside other data-crunching semiconductors.

But as chipmakers plough resources into lucrative HBM, shortages of the less flashy memory chips in everyday consumer electronics are pushing up prices, with Apple hiking the cost of its MacBooks and iPads.

Counterpoint Research analyst MS Hwang said SK hynix wants to triumph over Samsung in the red-hot memory chip market.

“Along with the HBM leadership it has demonstrated until recently, the company is now planning to take the lead in terms of volume as well,” Hwang told AFP.

“The use of funds from its US listing can support such a goal,” Hwang said.

The AI chip boom has fuelled debate over what South Korea should do with the tax windfall, as well as workers’ demands over pay packages — with Samsung averting a strike by agreeing a deal on bonuses.

An image of an SK hynix jacket also went viral online this year as a symbol of wealth and success, with parody posts depicting it as a “golden ticket” to luxury boutiques or better dating prospects.

“SK shares the understanding that effective employee compensation and better market capitalisation are essential conditions,” Counterpoint’s Hwang said.

SK hynix has said proceeds from the offering will be used to fund new chip factories to ease shortages.

The company, along with Samsung, is involved in a massive public-private investment of 800 trillion won to build a new semiconductor fabrication hub in southwest South Korea.
EU parliament greenlights digital euro

AFP
July 9, 2026 

If all goes smoothly, the EU hopes to have the digital euro up and running in 2029 – Copyright AFP Nicolas TUCAT


EU lawmakers on Thursday backed kickstarting negotiations on creating a digital version of the euro, as the European Union races to introduce the currency.

The project has been years in the making but the EU first needs to agree on the legal framework underpinning the virtual currency before it is available for use.

The EU believes a digital euro is the answer to cutting its addiction to US payment systems like Visa and Mastercard as well as Apple Pay and Google Pay.

A European Parliament committee had already given its green light but that approval was challenged and there was a call for an assembly vote.

An overwhelming majority of lawmakers approved establishing a digital euro.

That means negotiators from the parliament and EU capitals can start talks to reach a deal by the end of 2026. If they stick to that timeline, the European Central Bank hopes the digital euro would be available to citizens in 2029.

The first meeting of negotiators is expected this month.

There are also plans for a pilot programme in mid-2027 to test how the digital euro would work in practice if there is an agreement by the end of the year.

EU lawmaker Fernando Navarrete Rojas, one of the lead negotiators for the parliament, insisted the digital euro was “an alternative, not a requirement”.

He added that those saying the digital euro would be used “as a tool of control” were “lying”, stressing there would be the “highest privacy standards”.
How NATO leaders reacted to Erdogan’s revolver gift

AFP
July 9, 2026 

NATO leaders received the highly unusual gift of a revolver and six bullets
 – Copyright AFP SAUL LOEB


What does a world leader do with a gun and six bullets? That was the conundrum NATO leaders faced after the Turkish president offered them each a revolver after the Ankara summit.

British Prime Minister Keir Starmer was the first Wednesday to mention the highly unusual gift presented by President Recep Tayyip Erdogan to his guests.

On the flight back from Ankara — where NATO leaders had gathered for two days — Starmer said he and others received a revolver engraved with their name.

Alongside the gun sitting in a red box lined in black were six live rounds and a note exempting the weapons from export controls.

It was a surprising gift to say the least, several officials from the different alliance member states said, and gave rise to some “insane” scenes among the various delegations’ security teams.



– Headache –



Some leaders discovered their gift much later.

Belgian Prime Minister Bart De Wever only “learned of the exact nature of the gift” after landing in Belgium.

“The prime minister was surprised and immediately handed it over to airport police so it could be placed in a secure safe and the matter was handled in accordance with relevant procedures,” an official told AFP on Thursday.

De Wever’s security team also handled the revolvers given to EU chiefs based in Brussels, Ursula von der Leyen and Antonio Costa, with all the security and protocol-related headaches such an effort brings.

Von der Leyen “expressed her thanks” to Erdogan for the gift, her spokesman said, adding it would be decommissioned and donated to a military museum.

The revolver presented to Polish President Karol Nawrocki also arrived safely, but with the necessary precautions and a previous incident still fresh in everyone’s minds.

In December 2022, Poland’s police chief brought back an anti-tank grenade launcher from Ukraine that he had received as a gift.

The device exploded in his office, slightly injuring him and causing extensive damage to the police headquarters in Warsaw.

This time, “it is certain that no one is going to fire it,” an aide to Nawrocki told a local radio station.



– Why such a gift? –



Several revolvers, including those belonging to Starmer, German Chancellor Friedrich Merz and Dutch Prime Minister Rob Jetten, have for now remained in the Turkish capital.

And for good reason: depending on the laws in force, transporting firearms is often far from straightforward, especially when they are fully functional.

Canada’s Prime Minister Mark Carney took his revolver with him but left the ammunition in Turkey, Canadian officials said. They did not explain why.

The weapon given to Swedish Prime Minister Ulf Kristersson “will have to be transported to Sweden in accordance with all applicable procedures”, his team told AFP in a statement.

Beyond the logistical challenge, the gift also puzzled several delegations attending the summit, which focused on Ukraine, Iran, and relations with US President Donald Trump.

The question asked over and over again: why such a gift?

While it is very common for heads of state to exchange various gifts during meetings or summits, such exchanges rarely require these kinds of precautions.

Contacted by AFP, the Turkish presidency did not immediately respond.