Canva’s CTO encourages all 5,000 employees to find the AI tools that work best for them

“We want to have just a thousand blossoms bloom,” says Humphreys of this embrace of what he calls “permissive” licensing.

To be sure, Canva’s security and trust teams still vet all potential AI tools before employees are authorized to use them. And the company says it has rejected some vendors that didn’t provide clear assurances about data protections and at times, their cybersecurity practices.

But, Canva’s overall intention is to send a strong message to employees that they can play a big role in steering their own course forward with AI. Humphreys says that’s a rethinking of the more traditional approach, when IT departments would act as the custodian for all technology systems that would be deployed. He says now everyone is in research and development when it comes to AI.

“We want them to be intrinsically motivated to understand that their jobs have essentially changed,” says Humphreys.

That’s also where the great challenge lies for Humphreys and other C-suite leaders at Canva. They take on the responsibility of educating the workforce on how best to use AI to boost productivity.

In July, Canva will host an “AI discovery week,” which includes three days of learning courses and a two-day “hackathon,” giving employees a good amount of time from regular workday tasks to experiment with AI. The entire organization is invited to attend and the courses offered will be taught by Canva employees, including a mix of technologists and those who are more self-taught.

Every two weeks, Canva hosts an AI-focused virtual event that draws up to 2,000 attendees, where the company allows practitioners of AI to share their favorite use cases. One employee shared how they were able to create a custom GPT, which is a specialized version of OpenAI’s GPT models to perform specific actions, to help managers compare career targets with their notes about an employee’s performance. This has helped save time drafting the bi-annual performance management assessments at Canva.

In yet another example, engineers who have mastered AI coding tools like GitHub Copilot, Cursor, and Windsurf have been encouraged to show not only how they can prompt effectively with these tools but also display the limitations and where human input is needed. Their output is always reviewed by a peer, who is also encouraged to utilize AI for their own code comprehension.

Humphreys says new graduates have a disadvantage when it comes to coding, as they know how to use these tools but don’t have the on-the-ground experience to determine whether the AI-written code is good or bad. To that end, Canva is investing in internal mentoring with more senior employees and also created a custom GPT based on around 300 peer-reviewed articles written by engineers at Canva explaining how the organization conducts software development. Junior employees can ask the GPT questions the way they would a human peer.

“I think it’s a challenge for the industry and how we get junior engineers to make that transition to senior,” says Humphreys.

Humphreys says the “vast majority” of the company’s engineers have experimented with AI and 50% are using these tools daily. He’d like to see that number rise to 80% by the end of 2025. The productivity for software engineers is at around 30%, on par with what many other CTOs and chief information officers are seeing, but Humphreys says they don’t fixate too much on that. He’s really focused on adoption.

He joined Canva in 2014 as a senior engineer and rose up the ranks several times before being elevated to the role of CTO in September. Humphreys oversees all technology, including IT, product development, infrastructure, security, and data science. Externally, recent launches have included Canva Sheets, an AI-powered collaborative spreadsheet product, and AI video generation with audio, which is powered by Google’s Veo 3 model. The company says Canva AI now generates eight-second video slips with synchronized audio.

For the internal applications of AI, Humphreys stresses the importance of keeping humans in the loop, whether AI is being used to help source code, write a press release, prepare a job assessment for a colleague, or help formulate a marketing plan. Employees will always be judged for all of their AI-assisted work.

“Humans are ultimately the owners,” says Humphreys. “They really have to own that output.”

John Kell

Send thoughts or suggestions to CIO Intelligence here.

NEWS PACKETS

Meta is hiring top AI talent, but skepticism on strategy is swirling. Mark Zuckerberg is spending big bucks to bulk up in AI, including a $14.3 billion deal to invest in Scale AI and hire cofounder Alexandr Wang, exploring takeovers of firms like video startup Runway AI, and efforts to bring on Safe Superintelligence CEO and founder Daniel Gross and former GitHub CEO Nat Friedman. But as Bloomberg notes, Meta insiders think the company’s AI strategy isn’t focused, that external applications have been disappointing, and that there’s been notable churn in leadership at Meta’s AI divisions, including the departure of most of the employees that built Llama, Meta’s large language model. “Llama 4 that came out this year was an absolute failure,” D.A. Davidson analyst Gil Luria told CNBC on Friday. “Meta can’t afford to fail in having the leading AI model. So they’re out in the marketplace desperately trying to replace their AI team right now.”

Aflac is the third insurer hit by a cyberattack this month. Aflac, which provides home and life insurance for more than 50 million policyholders, disclosed last week that cybercriminals were able to breach the company’s U.S. network and may have accessed customers’ personal data including Social Security numbers. Aflac says it identified the bad actors within its U.S. network on June 12 and believes it stopped the intrusion within hours. The company also says it believes that this attack is part of a broader campaign against the insurance industry and to that end, two other insurers—Erie Insurance and Philadelphia Insurance Companies—also disclosed breaches in June. Beyond insurance, cyber is also a rising fear for federal officials who have warned attacks may grow amid the escalating tensions between the U.S. and Iran.

Recruiters are getting buried under a pile of growing AI-enabled job applications. Anxiety has been growing about the impact of AI on white-collar workers, especially following Amazon CEO Andy Jassy’s comments about the company’s future headcount needs and Sam Altman saying AI can now rival a PhD on some tasks. But what about the actual impact of AI on hiring practices? The New York Times does a deep dive, which reports that the number of applications submitted on LinkedIn has increased 45% in the past year. The platform averages 11,000 applications per minute, and generative AI tools are contributing to this massive dump of content that makes it harder for recruiters to determine which résumés—or even the applicants themselves—are real. 

Former OpenAI CTO’s AI startup raises $2 billion. Thinking Machines Labs, an AI startup founded in February by OpenAI’s former CTO Mira Murati, has officially closed on a seed round that values the company at $10 billion, according to the Financial Times. Not much is known about Thinking Machines, beyond some vague language on the website including saying it intends to “make AI systems more widely understood, customizable and generally capable.” The company is packed with AI talent from OpenAI and the big fundraising round vaults Thinking Machines to a select—but growing—group of over two dozen U.S. AI startups that have already raised $100 million or more in 2025.

ADOPTION CURVE

Data concerns are the top reason why the majority of AI projects aren’t on time. A survey of 1,000 business decision-makers found that 69% of organizations report that at least one of their planned AI initiatives is behind schedule. The top three leading causes behind these AI project delays were concerns about data quality and availability (39%), challenges integrating with legacy systems (37%), and regulatory uncertainty and compliance challenges (36%).

The findings, conducted by Zogby Analytics in May and backed by AI management company Prove AI, also found that while 86% of companies are taking an active role in training and fine-tuning custom AI models, 55% say this process has been more difficult than they anticipated.

A vast majority of organizations polled (81%) say their annual AI budget is at least $1 million, with one out of every four companies investing $10 million or more.

JOBS RADAR

Hiring:

- Candid Health is seeking a principal software engineer, based in New York City. Posted salary range: $238K-$322K/year.

- Zip Co is seeking a VP of U.S. engineering, based in New York City. Posted salary range: $240K-$300K/year.

- ICapital is seeking a technical product manager, SVP, based in New York City. Posted salary range: $200K-$240K/year.

- Grant Thornton is seeking a head of AI, based in New York City. Posted salary range: $206.3K-$395.4K/year.

Hired:

- McDonald’s (No. 165 on the Fortune 500) promoted Valerie Ashbaugh to serve as U.S. CIO, after she spent over four years leading the restaurant chain’s global commercial products and platforms team. Before joining McDonald’s in 2021, Ashbaugh spent over eight years in various leadership roles at Nike, including as SVP of product, planning, and manufacturing. She also previously worked at Boeing, Dell, and Accenture.

Every Friday morning, the weekly Fortune 500 Power Moves column tracks Fortune 500 companies C-suite shifts—see the most recent edition.

- Legends named Jim Scholefield as CTO, where he will oversee the live events company’s technology strategy, innovation, and execution across all digital platforms and systems. Scholefield joins Legends from hotel chain Marriott, where he served as chief information and digital officer. He has also held leadership roles at Merck & Co., Nike, and Coca-Cola.

- Love’s Travel Shops announced the appointment of Tim Langley-Hawthorne as CTO, effective immediately, where he will drive product development and the deployment of AI. Before joining the truck stop operator, Langley-Hawthorne served as CIO at car rental company Hertz and held executive technology leadership roles at Hitachi Vantara, Western Union, and Information Services Group.

- Massachusetts Institute of Technology appointed Gaspare LoDuca as CIO and VP for information systems and technology, effective August 18. LoDuca joins MIT after serving as VP of IT at peer Columbia University. At MIT, LoDuca will oversee IT infrastructure and services to support the research and academic enterprise across student and administrative systems, cloud services, cybersecurity, and customer support.

- Provi named Qingqing Ouyang as chief product and technology officer, where she will lead investments in digital innovation for the online business-to-business marketplace that connects alcohol suppliers, distributors, and retailers. Most recently, Ouyang served as CPTO at real estate software company OJO Labs and also held leadership roles at tech giants Dell and Oracle.

- The Denver Broncos have hired Daniel Brusilovsky as CTO, joining the NFL team after previously serving as VP of technology at the NBA’s Golden State Warriors. Brusilovsky will oversee all technology services and systems for the team at the team’s training facility Broncos Park and the stadium Empower Field at Mile High.

- FloQast announced the appointment of Sri Ramalingam as CTO, joining the accounting software provider to advance AI-powered automation capabilities, including the recently launched FloQast AI Agents. Ramalingam most recently served as SVP of engineering at software developer Harness. He previously served as VP of engineering at Zoom and also held leadership roles at Cisco and Plantronics.