 |
| A composite image shows Phia co-founder Phoebe Gates alongside a promotional image for the AI-powered shopping startup. Phia is facing scrutiny over its affiliate sales practices and how it attributes online purchases. |
Phia, the AI shopping browser extension co-founded by Sophia Kianni along with Phoebe Gates, daughter of Bill Gates and Melinda French Gates, is facing scrutiny over how its technology handled affiliate sales. Questions about Phoebe's technical background have sparked controversy, but the more consequential issues concern Phia’s software and its sales attribution system.
Launched as an AI shopping assistant, Phia helps consumers find lower prices online, and the company has advertised its ability to search more than 220,000 websites.
Reports in recent months have raised questions about whether the company’s software improperly claimed credit for sales generated through other affiliates. If so, Phia could have received commissions on transactions it did not drive.
The controversy has developed alongside questions about Gates’ education and computer science experience.
A Page Six report published Aug. 14 cited a Stanford University classmate who questioned Gates’ technical background and said she did not take advanced coding courses at Stanford.
Gates graduated from Stanford in 2024 with a bachelor’s degree in human biology after completing the degree in three years. Stanford describes human biology as an interdisciplinary program that can include coursework in areas such as natural sciences, mathematics, computer science and engineering. However, which courses a student takes depend on their individual academic plans.
The fact that Gates did not take advanced computer science courses, however, does not by itself establish whether she has the technical skills needed to help run a technology company.
Startup founders also do not necessarily need to be advanced programmers. Technology companies are commonly built by teams in which founders bring expertise in areas such as business, product development, research or marketing while engineers develop and maintain the software.
That makes the affiliate controversy more important than the question of how much code Gates has written.
How Phia’s affiliate system came under scrutiny
In July 2026, the Phia app was accused of fraudulently and deceptively taking commission for fake clicks and sales it did not drive (a practice known as "cookie stuffing") and was suspended from the leading affiliate and influencer platform Impact.com.
An investigation by Bloomberg News, alongside findings from independent consultant Ben Edelman, found that Phia would secretly open a new tab in the background whenever a user shopped at an online retailer, even if the user arrived directly or through another affiliate program.
Edelman's analysis specifically identified this behavior in the iPhone version of the service, noting this feature was present in a release dated December 13, 2025. He also provided evidence of the same practice continuing in later versions.
He rejected Phia leadership's claim that this was a "bug," stating that the deliberate creation and deployment of a feature called "auto_drop" for automatically setting affiliate tracking "cannot be dismissed as a mere bug. It's intentional, and it violates both trust and contract.
Those findings came from Edelman’s own testing, technical analysis and code review.
In August 2026, Bloomberg further reported, citing leaked Slack messages and people familiar with the matter, that co-founders Gates and Kianni had been aware of the cookie-stuffing practice since at least December 2025 and had discussed it with executives and engineers.
Investors have put substantial money into Phia
The controversy comes as Phia has attracted significant investor attention.
Fortune reported that the company had raised about $43.5 million in total funding by 2026 and was valued at approximately $185 million. In May, its most recent financing round reportedly brought in about $35.5 million.
Affiliate revenue is particularly relevant to Phia because commissions from purchases are an important part of its business model. If transactions were incorrectly attributed, the financial impact could extend beyond individual commissions and raise questions about the reliability of the company's revenue reporting.
Reports have cited different figures for Phia's daily revenue and the financial impact of the disputed attribution system; these figures vary and should not be treated as definitive measures of revenue or losses without supporting documentation.
Phia has said it is reviewing transactions with affected brands and working to correct improperly attributed commissions. The company has also said it plans to appoint a chief compliance officer to strengthen oversight.
What the legal questions do and do not show
The controversy has prompted discussion of possible violations of U.S. federal fraud laws, including the federal wire fraud statute, Section 1343 of Title 18 of the U.S. Code. The statute provides for penalties of up to 20 years in prison in qualifying cases.
That does not mean Phia's conduct has been established as wire fraud. A criminal wire fraud prosecution generally requires proof of specific elements, including a fraudulent scheme and knowing and intentional participation in that scheme. A software error does not, on its own, establish criminal intent.
For that reason, the allegations reported so far are not enough to conclude that Gates or Kianni face a potential 20-year prison sentence. The legal question would turn on evidence about what happened, who knew about it, when they knew, and what they did in response. Those are questions that cannot be answered simply by pointing to the software's behavior alone.
Gates’ family name adds another dimension
As the daughter of Bill Gates, one of the world's best-known technology entrepreneurs, and Melinda French Gates, a prominent philanthropist, she entered the startup world with a level of public attention that few founders experience.
In an interview with Fortune, Gates said she wanted to build Phia independently of her family's name and influence. She also said her parents did not invest in the company.
That background does not establish anything about Phia's technology or business practices. It does, however, help explain why questions about the startup and its founder have attracted unusually broad attention.
The focus on Gates' computer science coursework may be more publicly visible than the technical dispute itself. Yet the more consequential questions are about how Phia's affiliate system was designed, tested, and monitored.
The questions that matter most
The central issue is not whether Gates could personally build Phia's software from scratch. It is whether the company had adequate controls around a system that could affect how affiliate sales were credited.
That includes how the technology was tested, how attribution was monitored, and what safeguards were in place to prevent Phia from receiving credit for a sale generated by another publisher or affiliate.
The timing of the company's knowledge also matters. If internal records show that employees identified the behavior months before it became public, investigators and business partners could reasonably ask how the issue was handled during that period.
Those questions will require evidence. Technical testing can establish what the software did, transaction records can show how sales were attributed, and internal communications can help establish what employees knew and when. Regulatory or legal proceedings, if any, could provide a separate assessment of whether the conduct violated applicable law.
The available information does not establish that Gates lacked the technical ability to lead a technology startup, nor does it establish that Phia's affiliate system amounted to criminal fraud. It does, however, show why the company is facing scrutiny.
For Phia, the important test is whether it can account for how its technology attributed sales, explain when the disputed behavior was identified, and demonstrate that its affiliate practices can be independently examined and verified.
