Digital transformation, one smart step at a time

Finance transformation will be a major priority for Hewlett Packard Enterprise CFO Marie Myers this year as her team looks to do more with artificial intelligence agents, focusing on use cases such as forecasting and accounts receivable processing, the executive said in a CFO Dive interview.

The team is scaling up its use of “Alfred,” an internal agentic AI tool that it developed with engineers at Big Four accounting and consulting firm Deloitte with the goal of boosting finance operations, Myers said.

After successfully piloting the tool last year, the company is now poised in 2026 to drive “far more uses of agentic AI inside of finance,” she said.

The agentic AI push reflects a rising trend among corporate finance teams, according to Deloitte’s latest CFO Signals Survey released in January.

More than half (54%) of surveyed finance chiefs said integrating AI agents in their departments will be a digital transformation priority in 2026. “This likely reflects growing interest in advanced tools to augment workflows and decision-making,” the report said. 

For her part, Myers has embraced a “transformation with purpose” mindset, with a focus on using forward-looking, enterprise intelligence to drive growth, HPE and Deloitte said Thursday in a blog post highlighting their joint finance initiative.

Myers became CFO of HPE in 2024 after a three-year stint as finance chief of HP Inc. She previously held a variety of other positions at HP, including chief transformation officer, overseeing the company’s information technology and transformation organizations.

In her current capacity, Myers has prioritized greater technology adoption at HPE, according to the blog post. Beginning in 2025, she launched an effort with Deloitte to transform HPE’s finance operations “and pioneer agentic solutions that can reimagine how CFOs lead their organizations,” the post said.

AI investment race

HPE’s intensified use of Alfred comes as technology giants and Big Four firms are heavily investing in AI to beef up their offerings while experimenting with use cases internally.

Deloitte announced in March 2025 that it was launching a new suite of “ready-to-deploy” AI agents, dubbed Zora AI and built on technology from NVIDIA, for automating workflows across business functions, including financial management. The consulting firm envisioned its new agents being used to complete complex tasks and provide “on-demand insights, analysis, reporting, workflow automation, decision support and data sourcing,” according to a press release at the time.

As part of last year’s announcement, Deloitte said it was collaborating with HPE to offer a finance-specific Zora product to customers of HPE Private Cloud AI, with both companies planning to use the tool internally in the meantime.

The jointly developed finance tool, which is now referred to as CFO Insights, is cutting HPE’s financial reporting cycle by about 40%, while driving “more productive and targeted discussion around operational performance,” according to the Thursday blog post.

“By embedding agentic AI and generative AI into finance, HPE is demonstrating an approach many teams are still working toward: reimagining processes, moving pilots into production at scale, and generating real, measurable value,” Abdi Goodarzi, U.S. chief commercial officer for Zora AI at Deloitte, said in an emailed statement.

HPE affectionately named its own internal version of the tool “Alfred” after superhero Batman’s trusted butler, according to the post. Myers said Alfred is a platform that leverages both generative and agentic AI, bringing together data across key aspects of the company’s finance supply chain.

The tool allows people in finance and other parts of the company “to really interpret data and understand not just what is happening, but why it’s happening,” she said. “So, it’s a real time sort of interface for us on the key stats and performance of the company.”

‘Deterministic outcomes’

HPE worked with Deloitte and NVIDIA to ensure “deterministic outcomes” with Alfred, “meaning users get the same answer every time they ask the same question, according to the blog post.

“That might sound simple, but it solves a long-standing theoretical challenge in enterprise generative AI: making large-language models behave predictably in inherently non-repeatable LLM [large language model] environments,” the post said.

Myers’ deployment of Alfred started last year with a use case involving HPE’s regular operational performance review process.

“We chose really consciously to pick one of the most strategic processes in the company, which is our weekly ops call” that includes people from both finance and sales, Myers told CFO Dive. “Once a week, we come together on a Monday with 40 to 50 people who sort of really look at what’s happening with the business, and that call is the heartbeat of the company.”

Before the deployment of Alfred, “it was a very manual-driven, rear-vision mirror look-back call. We would spend literally days beforehand pulling together hundreds of PowerPoint pages … on what had happened in different aspects of the business.”

Alfred is capable of instantly analyzing data and generating insights, eliminating the need for tedious manual work and allowing the focus to shift from figuring out what happened to a more forward-looking conversation on “what we plan to do about it,” Myers said.

AI agents “now perform a lot of the analytical work that analysts were doing before,” she said. “So, analysts who were previously doing calculations around shipment data, shipment accuracy — those tasks are now performed by agents.”

Myers said the ops project gave her team “the chance to really position ourselves as finance helping to inspire and lead a broader enterprise transformation.”

She said the company is now looking to deploy Alfred on the “transactional side of finance” — from credit and collections to accounts payable and receivable processing.

“We’re in production right now with Deloitte on many of those initiatives,” the finance chief said. Forecasting could be the next major focus, she said.

For SAP Americas CFO Sonja Simon, artificial intelligence and its potential to change the way we work represents “a great opportunity to help get the finance organization ready and excited about what will come,” she told CFO Dive.

AI very much remains a “fast-moving train,” she said, with the technology continuing to evolve at a rapid pace and with much of its potential still left unclear or untapped.

When “there is such a dramatic change in what technology has to offer, it is only human to initially say, ‘Wow, what does it mean for me? How is it going to impact my job? Will I still have a job tomorrow? And if I have a job, then what will it look like?’” Simon said in an interview. “I feel, as a leader in finance, we really have to help the teams” work through that response, she said.

Getting comfortable

Strategically leveraging technology, including automation, is a critical part of how the Walldorf, Germany-based company approaches running finance. Over the course of her 18 years at the multinational software provider, SAP has grouped the tasks of its finance function by creating dedicated teams for different areas, such as human resources, compensation and revenue accounting.

The move has allowed SAP to become “very, very lean,” and to leverage technology: the finance team Simon oversees in her current role is comprised of 30 people and manages both the revenue and associated expenses of approximately 40% of SAP’s global revenue, she said.

Simon has held various roles during her near-two decade career with SAP. She was previously responsible for external reporting before being appointed to the role of CFO for Latin America and the Caribbean in 2020. She took her current role as CFO of SAP Americas in April 2025, which added the U.S. and Canada to her remit, she said. Prior to SAP, she worked for seven years at the Big Four firm Ernst & Young.

As AI and automation continue to evolve, it’s important for the finance function to become familiar with new tools. Within Simon’s team, “there are some small things that we do to ensure that people get comfortable” with AI, she said. That includes bringing AI up in team meetings  and asking for volunteers who are willing to talk about how they have used the technology lately, she said — whether in a professional or personal setting.

Simon also has an individual on her team that is “tasked with continually working on what is available for us within finance at SAP today, already, to ensure that we are really at the forefront of the adoption,” she said. As a global entity, there are numerous teams at SAP which are all examining how AI could be tapped within their functions, and it’s important that finance stays top of mind.

“We always want to make sure that they don't forget about us, so that…they ask us, ‘Okay, do you want to be the first user? Do you want to be the early adopter?’” she said.

Creating a ‘failure culture’

To drive early adoption of emerging tools and technology, however, it’s key to create the room for professionals to experiment and make mistakes. Simon said she is “a strong believer in creating a failure culture,” which can particularly be a challenge in finance. As the numbers crunched by the function make their way into the reports that govern executive decisions on, there’’s little room for error.

However, opening up space for financial professionals to experiment with new and emerging tools, and to make mistakes, is also crucial to driving creativity and finding new solutions, she said.

“The tricky thing here is you really have to celebrate failure, and you really have to make it stick, because it is only human,” Simon said of her approach.

For instance, Simon challenged the team at SAP to find better insight per deal for a key performance metric at the company: current cloud backlog, a KPI which is relied upon by investors and analysts to evaluate its cloud journey.

The team did come up with a tool that helps to simulate the impact on that metric, she said.While that tool may not give the right figure 100% of the time, “that is an example of where I say, ‘Okay, what you have put together is a lot better than what we had, which was, in that case, nothing,’” she said. “’And so the fact that this isn't 100% correct, keep in mind that you still improved the process significantly.’”

Sponsored content

By GEP

• The deployment of AI technology has created a cultural gap in many organizations.
• Business leaders must take steps to fill this gap by encouraging a culture of learning and experimentation.
• This will allow people at all levels to make the best use of powerful new systems and tools.

Why are so many businesses still waiting to deploy AI in their operations when the value it delivers is undeniable? Are they witnessing a readiness gap within their workforce?

As AI technologies mature and their use cases within procurement and supply chain operations expand, leaders need to have a plan to align their people and processes with technology. Their tech stack may look very impressive with state-of-the-art tools and features. But their efforts toward digital transformation can be futile if teams don’t understand how to leverage technology or feel threatened by it. 

If your employees are shying away from AI conversations, it’s a clear sign that they are not ready. What should you do to prepare them? 

You need to build an AI culture.

The question is: How do you create such a culture? 

Is Your Business Ready to Embrace AI?

Learn How to Identify People and Process Gaps

The Need for Cultural Adaptation and Integration

Many businesses now have sophisticated AI tools, but only a few have learned to use them well. These businesses have prioritized AI literacy, curiosity and judgement that help people make sense of powerful new systems and tools.

Cultural adaptation and integration are the most important factors that will help organizations realize the true benefits of AI technology, states the GEP Outlook Report 2026. It involves creating user comfort, encouraging experimentation and streamlining decision-making and interaction between humans and machines. Companies will also need to redefine the rules of engagement between humans and machines within an integrated system.

Business leaders should not assume that their AI investments will naturally lead employees to gain new capabilities and skills. Instead, they must create an environment that encourages learning, even if that means making mistakes initially.

Check Out GEP’s AI Powered Procurement Software

How to Build an AI Culture

Three elements are essential to building an AI culture, notes the GEP Outlook Report 2026. These are:

Provide AI literacy

Leaders must support AI literacy at all levels. Category managers, logistics analysts and supplier relationship leads should all learn how to use AI tools in routine operations. They should also understand how AI models are trained, what data biases are, what to look out for and when to escalate issues. 

Define new rules of engagement

Leaders should use process management tools and identify the roles and responsibilities of all stakeholders. This will provide more clarity and remove ambiguity about who does what. It will also create accountability for errors or sub-optimal choices. 

Don’t worry about outcomes

Employees may take a while to start feeling comfortable working with AI tools. Leaders can shorten this learning curve by building a culture of experimentation. Encourage teams to become familiar with new tools by testing and experimenting with them. When employees aren’t afraid to experiment, they are more likely to adapt and adjust quickly.   

Encourage Curiosity and Continuous Learning

AI models can change quickly. When models change, workflows may change as well. This makes it imperative for organizations to invest in continuous learning. Your employees should always be eager to explore new ideas, ask questions and seek out solutions. When you build the right culture, people are more inclined to explore the how and why behind things.

Leaders should invest in training programs, online courses and certifications in AI and machine learning. This will help people at all levels to gain a basic understanding of AI.

Leaders can also invest in workshops to stimulate creative thinking and innovation. In a learning environment, employees will be comfortable brainstorming ideas and exploring AI’s potential use cases.

From Tasks and Workflows to Agentic Orchestration

Learn How GEP Qi is Redefining the Way Procurement Works

Build a Digital Workforce

There is little doubt that businesses need to leverage AI technology and tools. But as they deploy these advanced systems, they also need to build a workforce that is skilled and AI-ready. Neglecting this will create a gap between planning and execution.

You don’t want a situation where your employees ignore new tools or simply accept AI output without any judgment or oversight.

A culture of learning will allow employees to become familiar with new systems and tools. It will help them learn at their own pace. Over time, they will start trusting these tools, understand how to interpret output and get comfortable with new roles and ways of working.

In the race to digitize credit risk assessment, one mistake keeps tripping businesses up: chasing shiny new tools before fixing their data foundations. The truth is: no technology can outperform poor data. And when operating across borders, that failure is magnified.

For CFOs, the stakes are high. They’re tasked with managing bad debt, driving cost savings, optimizing customer experience, and preventing fraud, all while credit risk grows more complex.

According to Dun & Bradstreet’s latest Global Business Optimism Insights Report, financial confidence fell 2.3% globally in Q4, driven by weaker demand and regulatory uncertainty. This signals growing caution among finance leaders, reinforcing the need for smarter, data-driven credit risk strategies.

So, how do CFOs move from ambition to action? The path begins with a 'data-first' approach to credit risk management.

Digital innovation is more than just risk control; it's a powerful business enabler. Leaders need to define what success looks like and ensure it can be tracked. That means asking: What are my current credit risk capabilities? What do I want them to be? What are the core drivers behind the project?

Improving cash flow, preventing fraud, and reducing Days Sales Outstanding are measurable outcomes that matter. Key Performance Indicators can cover anything from bad debt and data quality to customer satisfaction and process efficiency. Whatever the metrics, they should be revisited regularly to ensure the sustained value.

Digital transformation starts with data hygiene, and that challenge multiplies across borders. For global CFOs, the data needed to accurately assess international credit risk is often messy and inconsistent, varying in format, language and convention. This chaos renders it unusable for a centralized, intelligent risk engine, leading to fragmented insights and poor decisions.

To fix it, CFOs must audit for financial inconsistencies and cleanse historical records; enforce data quality at the source to prevent future problems; and standardize formats while establishing a universal, common data taxonomy across all regions and subsidiaries. This turns fragmented local data into consistent, usable, global credit risk intelligence.

Next, enrich and automate this standardized data using third-party sources, and govern it carefully for compliance across all international jurisdictions. Without this foundation, even the most advanced AI and machine learning tools will fail to deliver consistent, predictable results.

Transformation is a chance to rethink credit review workflows — not replicate them. This requires a willingness to evaluate and redesign processes addressing deep-seated issues like poor team coordination (e.g., between credit, finance and customer service). The goal? Leaner, smarter processes that align with modern systems and customer needs.

To gain a truly global view of credit risk, CFOs must eliminate silos by connecting data across their disparate systems, creating a unified and consistent view of risk. This consolidates risk-related information into a single source of truth, regardless of its country of origin. Running systems in parallel can help map existing decision-making and effectively tune new credit policies before fully going live.

Time, people and training

New software may roll out fast, but people need time to adapt. Success depends on planning for what happens before, during and after implementation.

Achieving stakeholder buy-in is crucial, which means defining clear roles, managing expectations, and empowering teams to lead improvements. When stakeholders can link credit KPIs to wider business goals, they’re more likely to embrace new tools.

Don’t skimp on training. Even the most effective tools can stall without ongoing support. Training must be comprehensive, continuous and focused on how new tools turn data into smarter credit decisions.

Consistent, structured, and, crucially, globally standardized data is the core of successful credit risk transformation. Without it, even the most advanced AI or analytics tools become unreliable.

Poor data hygiene and a lack of standardization across international operations lead directly to bad decisions, costly compliance issues, and wasted investment. The journey to digital maturity in finance is not about buying the fastest engine; it’s about building a consistent roadmap and ensuring the quality of the fuel that drives it.

U.S. companies are investing more heavily in emerging technologies such as artificial intelligence than their global counterparts, spending an average of $190 million annually compared to the global average of $174 million, KPMG said in a recent report.

American firms reported seeing financial returns of $293 million on average in the past 12 months, higher than the $265 million global average, according to the findings released last month. Despite out-investing their global peers and seeing higher financial returns, U.S. companies are struggling to achieve full maturity in their emerging technology strategies due to hurdles such as cost concerns, the Big Four accounting and consulting firm said.

“The survey data says that scaling emerging technology investments is taking longer than initially anticipated,” Marcus Murph, head of technology consulting at KPMG US, said in a press release.

On average, organizations globally expect to allocate 5% of their annual budget to AI initiatives in 2026, up from 3% in 2025, Paris-based technology consulting firm Capgemini said in a January report.

“As AI moves from experimentation to enterprise-scale deployment, organizations face a new challenge: translating rapid advances in AI capabilities into sustained business impact,” the report said. Over the next 12 months, organizations plan to accelerate investments in infrastructure, data, governance and workforce upskilling, with the goal of laying a strong foundation for “sustainable AI adoption,” according to the research.

KPMG found that U.S. companies in particular are planning this year to outspend global counterparts across various technology categories including AI, cybersecurity, data/analytics and “post-quantum cryptography,” which refers to new types of digital encryption designed to protect data from being cracked by future quantum computers.

While U.S. companies are scaling a wide range of emerging technologies, few have reached full operational maturity, with only 10% describing their tech implementations as “fully scaled and continually evolving,” down from 25% last year, the Big Four firm reported. 

More than half (56%) of respondents told KPMG the cost of fixing technical debt prevented them from making investments in new technology programs.

The results show that companies are finding technology projects to be much more complex than anticipated, especially in areas such as data modernization, security and accuracy, according to Gary Plotkin, digital platforms lead at KPMG U.S.  

“This complexity has led to cautious adoption, particularly within core business functions, as companies are unsure if the technology is mature enough for large-scale deployment,” Plotkin said in KPMG’s release.

The rising use of artificial intelligence tools in investor research is leading to heavier demands on CFOs and their teams, according to Gartner researchers.

Of 146 CFOs surveyed by Gartner from October through December, 43% said they were spending more time preparing for earnings calls, the Stamford, Connecticut-based research and advisory firm said last week in a news release summarizing a December client report. Another 34% of respondents cited an increase in unexpected investor questions during earnings calls, while 38% reported more media and social media engagements.

“It’s coming from an investor base that is more informed thanks to AI,” Dymah Paige, a senior research analyst in Gartner’s finance practice, said in an interview. “CFOs will need to leverage AI tools themselves to react in a more targeted and timely manner.”

Rapid advancements in AI, combined with other factors like persistent economic volatility, are causing the CFO’s role to expand at an unprecedented rate, according to Bruno J. Navarro, a finance thought leader at Pleasanton, California-based software giant Workday.

“The CFO of 2026 (and beyond) won’t just manage finances — they’ll be a strategic navigator, co-piloting the business alongside the CEO,” Navarro said in an Oct. 17 blog post.

This shift, he added, will require a “fundamental change” in how CFOs operate, pushing them to focus on priorities like leveraging AI and data to deliver a holistic view of organizational performance.

“This leads to the rise of financial storytelling, a critical skill for the modern finance leader,” he said. “It’s the ability to translate complex data and financial models into a clear, compelling narrative that informs and influences the C-suite, the board and investors.”

Today, CFOs are asked to deliver more, with unpredictable markets only heightening the pressure, according to Gartner’s report, which was shared with CFO Dive. “The rising volume of investor interactions is only one symptom of this,” it said.

A 2024 survey by investor relations advisory firm Corbin Advisors found that 58% of global institutional investors and securities analysts were using or evaluating AI tools in their investment and research processes.

This is leading to heightened expectations for a “responsive and informed management team,” Gartner said in its report.

Some of the world’s biggest companies are already using AI tools in their investor relations activities, according to Paige.

In one case study, highlighted by Gartner in a separate, Jan. 26 report shared with CFO Dive, Dell created a suite of AI capabilities to improve earnings call preparation, including a “sentiment analysis tool” for evaluating the alignment of messaging and business performance.

Dell didn’t immediately respond to a request for comment.

With artificial intelligence seeping further into everyday business processes and use cases, it’s critical for CFOs to pay close attention to the data feeding the AI models that help them scenario plan, create forecasts and develop long-term strategies.

In the age of artificial intelligence, CFOs should demand the same audit discipline for data that they do for financials, according to Donna Dror, CEO of Usercentrics.

“If you think of data as a financial asset, which I believe every CFO now does, then privacy and data trust are the governance mechanisms that protect and enhance that asset’s value,” she told CFO Dive in an interview.  

Developing a robust privacy infrastructure  

Data is a crucial area of focus for CFOs today as they integrate AI tools more widely into their companies, with many finance chiefs’ prioritiesdirectly tied to the technology. Finance chiefs today are excited about the many ways in which AI could potentially benefit their companies, whether it’s improving forecasting or helping to cut down on the time it takes for the financial close.

However, one of the barriers CFOs today find themselves facing is “that enthusiasm outpaces data maturity,” Dror said. “I think the biggest disconnect I see is confidence…because CFOs have to trust a financial statement, but they can't say the same about data feeding into, for example, AI tools.”

Dror has served in the top executive seat for the Munich-based consent management platform since October 2022, first joining Usercentrics as its chief revenue officer in 2021, according to her LinkedIn profile. Her past experiences include an eight-year tenure at fellow software provider Similarweb, and she has acted as an advisor with New York-based private equity and venture capital firm Full In Partners since May 2019.

Untrustworthy or unclear data can contribute to a number of operational issues inside of a business — inflating or distorting forecasts, for example, or exposing companies to both material and reputational risks, which all ultimately impact the bottom line, Dror said. For CFOs, that puts a premium on ensuring clean data: finance chiefs build strategies around “accuracy, predictability,” and risk mitigation, Dror said. “I’d say all three of those things really depend on whether your data is consented, complete and defensible.”

Every financial decision must also be traced back to clean, legitimate data, “so if I'm CFO, I'm going to view privacy infrastructure as part of the foundation of financial accuracy and brand resilience,” Dror said.

Compliance as a growth driver  

As finance chiefs take closer looks at their data, they need to be sure they are treating compliance as a core pillar, in a way that allows space for shifting regulations. AI privacy laws in the U.S. are still in their nascent phase both on a federal and state level, while global businesses need to grapple with a host of differing privacy and security laws across disparate markets.

The “truth of the matter is regulation always comes off the innovation, and we're not yet at the point where all of these data privacy laws are fully extending to every use case that now exists in the technological evolution that we're experiencing,” Dror said.

That means shifting how CFOs think about compliance — which, despite often being perceived as a barrier to expansion, can “fuel growth, Dror said. Keeping compliance in mind from the start can help businesses better create trust with their customers and build lasting relationships, for example, she said.

Additionally, “I think that the cost of retrofitting compliance later is always more expensive, so that's something that I think should be foundational to how these decisions are made now,” she said.  

Big Four accounting and consulting firm Deloitte’s Australian arm has come under public scrutiny after reports that it produced a document for the Australian government that was full of artificial intelligence-generated errors.

The incident involving one of the biggest names in the corporate finance world should serve as a wake-up call for anyone in the profession using AI tools without proper safeguards, experts say.

“This situation underscores a critical lesson for finance professionals: AI isn't a truth-teller; it’s a tool meant to provide answers that fit your questions,” Bryan Lapidus, FP&A Practice director for the Association for Financial Professionals, which provides training and credentials for people in the finance profession, said in an email.

Deloitte Australia’s report to the Australian Department of Employment and Workplace Relations was originally published on the department's website in July, according to an Associated Press article. A revised document was published earlier this month after revelations that the original version contained errors, including a fabricated quote from a federal court judgment and references to nonexistent academic research papers, AP reported.

Deloitte Australia agreed to partially refund the 440,000 Australian dollars ($290,000) paid by the Australian government for the document, which reviewed the information technology system used by the department to automate penalties in Australia’s welfare system.

“Deloitte conducted the independent assurance review and has confirmed some footnotes and references were incorrect,” a DEWR spokesperson said in an email, adding that “the substance” of the review has been retained. Deloitte has agreed to repay the final installment under its contract, the spokesperson said.

Deloitte Australia didn’t immediately respond to a request for comment.

Chris Rudge, a Sydney University researcher, has been credited with bringing the errors to light. He discovered that Deloitte’s report was “littered with citation errors,” prompting concern that it may have been generated by AI, according to an article published by the Australian Financial Review in August.

The event is a reminder for companies that AI is “smart but not infallible,” according to Nikki MacKenzie, an assistant professor in the Georgia Institute of Technology’s Scheller College of Business.

“We’re constantly hearing about how ‘intelligent’ AI has become, and that can lull people into trusting it too much,” she said in an email. “Whether consciously or not, we start to over-rely on it.”

AI tools such as generative AI chatbots can be susceptible to “hallucinations,” where they create outputs that are “nonsensical or altogether inaccurate,” according to an explainer published by ChatGPT creator OpenAI. Hallucinations can occur when an AI model is trained on a dataset comprising biased or unrepresentative data. AI tools can also be vulnerable to adversarial attack, where a bad actor manipulates output by subtly tweaking the input data, the article said.

Nearly six out of 10 employees admit to making mistakes in their work due to AI errors, according to a KPMG study released in April. In addition, about half use AI in the workplace without knowing whether it’s allowed and more than four in 10 are “knowingly using it improperly” at work, the study found.

The Deloitte event, while alarming, is unlikely to result in a pause in AI adoption, according to experts.

“I believe firms will see this as a normal cost of doing business,” MacKenzie said. “Just like how employees make mistakes, tools can too. The goal isn’t to avoid AI’s errors — it’s to make sure we’re smart enough to catch them as the ultimate decision-maker.” 

The Deloitte incident is just the latest high-profile example of the risks posed by AI as it gets rapidly deployed across industries.

In January, Apple suspended an AI feature designed to summarize news alerts after complaints that it generated false and inaccurate information. And in 2023, a New York federal judge sanctioned two lawyers who submitted a legal brief that included fictitious case citations generated by ChatGPT.

Jack Castonguay, an associate professor of accounting at Hofstra University, said Deloitte is the first major accounting firm he’s aware of that has produced a report with fictitious details generated by AI.

“It seems like it was only a matter of time,” he said in an email. “Candidly, I’m surprised it took this long for it to happen at one of the firms.”

The event underscores the need for organizations to build AI safeguards or internal controls to mitigate risks such as hallucinations, MacKenzie said.

“The responsibility still sits with the professional using it,” she said. “Accountants have to own the work, check the output, and apply their judgment rather than copy and paste whatever the system produces.”