Ethics in artificial intelligence is not a side project or a branding exercise; it is the operating system of the businesses that deploy it, Aarani Stoneham says.
From her seat as sole in-house counsel at ModiFace, the AR and AI subsidiary of L’Oréal Canada that powers virtual try-on tools for major beauty brands and platforms, she pulls legal risk, commercial pressure and ethical red lines into the same room. The Toronto startup, acquired by L’Oréal in 2018, builds highly proprietary algorithms that underpin white-label virtual try-on tools used by global cosmetics brands and big tech platforms. She says the principle is simple: If the technology touches consumers, you cannot outsource the moral responsibility.
That conviction did not appear overnight. Earlier, at the Law Office of Yaso Sinnadurai, she effectively ran a second office location, managing residential purchase, sale, and financing files from opening through to client and bank reports and final accounts. The role gave her operational autonomy and the confidence to later set up legal departments from scratch, and it hardened during an early in-house stint at a consumer financing company, where Stoneham was brought in to build the first legal department and establish governance across business units spanning sales, marketing, and HR. When the company’s model shifted, she chose to move on, and the next move pushed her into a very different corner of the corporate world.
At Whirlpool Canada, she was hired to set up a Canadian legal function within a multinational, where guidance from the US parent on legal processes was limited. That pushed her deeper into cross-border work and corporate strategy. She flew regularly to Benton Harbor, Michigan, to work with North American legal leadership, then translated global initiatives into contracts, governance frameworks and risk assessments that complied with Canadian law. That mix of autonomy and alignment set the stage for her current role.
It foreshadows what she is doing now at ModiFace. The L’Oréal-owned company builds proprietary, in-house virtual try-on technology and licenses it to internal and external beauty brands and tech platforms across dozens of global jurisdictions. Clients want aggressive protections such as sweeping indemnities, IP ownership concessions, and broad assurances that someone will take responsibility for issues or claims that arise, including end-consumer complaints and regulatory scrutiny. “When we have a prospective client, we negotiate the agreement, and our team helps them set up the tech, whether it is on their website or app. We even offer to provide them with sample requisite consent language,” she says. “But at the point that they deploy the technology, meaning they make it available to their consumers, and ‘go live,’ they are responsible for any legal obligations or issues that arise.”
That hard line on deployment is paired with an equally firm stance on ownership. The virtual try-on engine is built from scratch at ModiFace, and while big-name customers may push for joint IP or broad licences that erode control, Stoneham resists those pressures. “That’s ours. We cannot give them the rights to ModiFace’s proprietary technology,” she says, framing the company’s patents and trade secrets as an asset that must be tightly managed rather than traded away in negotiations.
Her docket is not limited to contracts. As the only in-house Canadian lawyer for a small but global-facing subsidiary, she drafts AI and data privacy policies that incorporate concepts from the GDPR and the EU AI Act, adapts L’Oréal standards to Canadian realities, manages corporate governance, and coordinates a lean external counsel panel covering IP, patents, employment matters and litigation. She describes herself as “very much a generalist,” but this plays out almost entirely in a high-velocity AI environment where the law is evolving, and regulators are watching.
The speed and complexity of that environment explain why ethics cannot be bolted on at the end. She works with developers and AI leads to understand how a model is trained before it reaches consumers and to reinforce the importance of datasets that are representative of individuals of all ethnicities, skin tones, and genders. The aim is to ensure that the resulting technology does not default to one consumer type and that bias is caught before it reaches the market.
“If I’m a 25-year-old developer from U of T, [bias] is not really at the forefront of my mind,” she says, since the developer is focused on the technology itself. “It is legal’s responsibility to advise other stakeholders across the organization that there are non-technology things they need to think about as they work on the technology.” Her goal is a partnership in which engineering talent, commercial teams, and legal constraints are pushing in the same direction, rather than legal acting as a late-stage brake on innovation.
The same instinct to build guardrails early, rather than clean up harm later, drives Stoneham’s work outside the office. She founded AI Literacy Learners after watching her five-year-old puzzle over “how a picture got on a screen” and after exposure to increasing news stories about young people whose interactions with chatbots ended in addiction, self-harm and death. She began questioning why children are being handed powerful AI tools before they understand, even in simple language, what those tools are, how they work and the risks they present.
AI Literacy Learners flips the usual narrative. Instead of racing to give students access to the latest education apps and essay generators, Stoneham wants schools to focus first on conceptual grounding, from kindergarten classes learning to identify AI, for example, distinguishing between machines that need human input and systems that act autonomously. Next, teenagers would be asked to critique generative tools that “make confident mistakes,” she says. She points to countries such as Finland, where children at age three are now taught to spot manipulated media and deepfakes, as evidence that Canada is already behind the curve.
For Stoneham, the appeal of ModiFace is that it allows her to embed that ethical framework at the source of decision-making inside a business built around its technology. She is a single lawyer at a relatively small AI company, but she sits alongside developers, data scientists, commercial teams, designers and product leads, shaping how proprietary algorithms are trained, licensed, and governed, and ensuring they are fit for the consumers who will eventually use them.
Stoneham says her ModiFace role is where she feels most successful, not because the legal challenges are any simpler than her previous roles, but because of the environment within which they unfold. “It’s the context I'm working within every day, the pace of the industry, the quality of the team members.” The work sits at the junction of AI, beauty and global consumer expectations, and her work goes beyond legal oversight; it is about ensuring that the company’s technology evolves with the same principles that guide its governance, so that the legal decisions it makes and the ethics it champions move in step with the technology ModiFace ultimately places in consumers’ hands.