AI killed Getty Images?

Getty Images filed for Chapter 11 in Delaware on Oct. 21. It is a prepackaged case, with terms agreed with creditors in advance, and according to the company’s announcement the business keeps running and should emerge as a private company around year-end. Getty family affiliates and Koch Equity Development are putting in up to $533 million of equity—not all of it new cash, since part of it converts existing debt and bankruptcy financing. If the plan goes through, today’s shares are canceled and nothing is distributed on them.

One outlet ran the news under the headline “AI Claims Another Victim.” To be fair, the rest of that headline lists debt and lawsuits alongside AI, and the article treats AI as one cause among several. But the culprit standing at the front of the headline was AI alone, and the headline is what people remember.

So the question “Did AI kill Getty?” should not be taken as asked. It looks like one question, but you cannot answer it until you decide how many boxes the cause of death gets. There is a form that has already drawn those boxes: the death certificate.

A Korean death certificate records the cause of death as a chain traced backward. The top line is the immediate cause; below it go the conditions that led to it, back to the underlying cause. Separate from that chain, a box at the bottom takes other conditions that affected the death without belonging to the chain. And among the most common errors, according to the Korean Medical Association, is writing “cardiopulmonary arrest” as the immediate cause. Everyone’s heart stops when they die. That is what death looks like, not why it happened (report on the association’s statement).

One thing first: Getty is not dead. Photos are still being licensed, and the company says it expects to keep meeting its obligations to contributors, photographers among them, in the ordinary course. What is receiving a certificate is the company that went public again in 2022—its stock. And in that death, “it could not pay” is the cardiac arrest. Being short of the means to pay is what a bankruptcy filing looks like, not why it happened. To fill in the boxes you have to look above it.

The materials here are the company’s own. I could not obtain the first-day declaration filed with the court; I worked instead from the disclosure statement Getty attached to its Securities and Exchange Commission filing the same day, and its August second-quarter results. Both are the debtor’s account, and that has to be carried all the way through.

Immediate cause: obligations the business could not earn its way out of. The disclosure statement puts annual debt service at the petition date at about $239 million, and says that burden, together with other pressures such as litigation and merger costs, exceeded the cash the business generates. The same document gives revenue for the latest twelve months as about $978 million. Just under a quarter of revenue was going to interest and principal every year.

Leading to it: a contract from 2020. When Getty went public again through a SPAC merger in 2022, it took over a warrant agreement the SPAC had signed in 2020—rights to buy shares at $11.50. When the stock briefly topped $30, the company blocked their exercise; investors sued, and Getty lost. By the disclosure statement’s figures, judgments and related interest exceed $200 million, defense costs over roughly four years reached $35 million, and the company estimates further exposure of more than $275 million from claims not yet brought—an estimate, not a debt already owed. It hit home in the second quarter. Free cash flow was negative $122.6 million, and the company attributed most of that to $110.9 million in judgment and interest payments. Nearly half a quarter’s revenue went out the door at once because of a contract another company signed six years earlier.

Alongside it: a merger abandoned. The $3.7 billion Shutterstock deal ended after Britain’s Competition and Markets Authority required the sale of Shutterstock’s editorial business as a condition of approval and Getty decided not to go through with that sale. The board chose to withdraw on June 30 (filing), and the agreement was terminated on July 7 (CMA). It was less a regulator blocking the deal than a buyer declining the regulator’s terms. Fees for pursuing it alone topped $100 million.

The bottom box: generative AI. Now AI. Its traces are real. Second-quarter revenue fell 2.5% from a year earlier, and the chief executive himself cited continued pressure in the agency business and iStock, the low-priced stock site. Over the twelve months to June, active annual subscribers fell from 321,000 to 240,000 and purchasing customers from 707,000 to 636,000. The disclosure statement names AI, along with changes in media consumption and marketing, as headwinds for iStock and agency customers. Then, in the same passage, it says those trends “would not alone have necessitated the commencement of these Chapter 11 Cases.”

Of course the company would say that. A debtor that has to sell itself to new investors is not going to write that AI finished its business. So I won’t lean on that sentence. Look instead at what the filings pin down and what they don’t. Nowhere in them is the revenue AI took away broken out; the company even notes that iStock’s drop includes ended free trials and partnerships and changes in search rankings. Nor is the 2.5% net decline a ceiling on AI’s share—growth elsewhere may have masked a bigger loss. The forces that pushed this filing, by contrast, come with dates and amounts on paper: debt service close to a quarter of revenue every year, a judgment payment worth nearly half a quarter’s sales, more than $100 million in merger fees.

The boxes on a death certificate exist to hold exactly this difference. The chain takes what follows from cause to effect; conditions that affected the death without belonging to that chain go in the bottom box. With Getty’s filings, only the debt side can be linked into a chain. So Getty’s certificate reads like this. Immediate cause: repayment obligations it could not meet. Beneath it: the inherited warrant contract and the cost of the abandoned merger. AI goes in the bottom box. In explaining this filing, I put the weight first on the obligations the documents capture. That is not proof that AI was nowhere in the chain. I put my confidence at about 7. The reason it is not 10 is plain: these filings cannot measure how much revenue AI actually took, or how much heavier that made the debt. If that number turns up and it is large, AI has to move up from the bottom box into the chain.

Turn the same numbers around and another picture appears. In the quarter that creative revenue fell 2.6%, editorial—news and sports photography—rose 9.2%. The company’s 10-Q attributes the increase mainly to sports, especially World Cup demand, and says the World Cup also tilted how subscription revenue was split between the two segments toward editorial. So the 9.2% is not evidence that photos became more valuable. An event happened, and people wanted pictures of it.

That is exactly the point, though. What grew was pictures proving a match really took place—pictures a generative model cannot make. The creative side includes staged images like smiling office workers in a meeting room, and those can now be generated in seconds. The filings do not say which kinds of images fell by how much, so this is not a tally; it is an illustration of how differently the two kinds of picture are used. My guess, held firmly in direction but loosely in size, is that what generative AI is doing to the photo market is less abolishing photographs than splitting them into decoration and evidence. One company and one quarter is a thin base for that.

Getty’s own fight with AI points the same way. It sued Stability AI in Britain, and in November 2025 the High Court dismissed its secondary copyright infringement claim, finding that the model itself was not an infringing copy of Getty’s photographs (judgment). The ruling did not decide whether training itself is lawful. In December, Getty was granted permission to appeal the dismissal of that secondary infringement claim (decision); I could not confirm what has happened in the appeal since. Either way, in Getty’s ledger of lawsuits, the one that actually tipped the company over was not against an AI company. It was against investors who wanted to buy its stock.

Why bother splitting this into boxes? Because the cause written in an obituary becomes the next patient’s prescription. If the story settles as “AI killed it,” the lesson left behind is “beware of AI.” Companies piling debt onto complicated listing structures then have no reason to read their own certificates, and AI gets credit for work it did not do and looks bigger than it is. Both are wrong maps. Read the certificate as filed and there are two lessons: it was the capital structure that pushed this filing first, and what AI is really changing is less whether photographs live or die than where their value attaches.

Getty, with less debt, plans to return as a private company at year-end. Whether it bets on decoration or on evidence will settle, after the fact, whether this was an AI story at all. This is one certificate filled in one way; anyone who would draw the boxes differently is welcome to.