Getty Images launched its own AI image generator in 2023. Three years later, on October 21, 2026, it filed for Chapter 11 bankruptcy as part of a planned restructuring.

An awkward example for anyone selling “adopt AI” as a survival plan.

The October 12 announcement includes up to $533 million in new equity and a planned return to private ownership. Getty expects to continue operating. Its debt and failed Shutterstock merger matter here, as does competition from AI-generated imagery. There’s enough going on without pretending a single technology explains the whole mess.

But the customer’s decision has changed. Someone who needs a generic image for a presentation can now generate one instead of buying it.

It might have strange fingers. The customer might use it anyway.

That’s uncomfortable for those of us who earn a living producing things. We can explain quality until we’re blue in the face. The person paying gets to decide how much quality the assignment requires, and sometimes the answer is considerably less than we’d like.

A photograph documenting an actual event has a different purpose. Its value depends on somebody having been there. But a stock image of smiling people in a meeting has a less comfortable future.

Chegg has already experienced what happens when customers find another way to get answers. Its second-quarter revenue fell 51% in 2026, after the company had identified AI and reduced Google traffic as major pressures. Stack Overflow’s chief executive described ChatGPT as an existential moment for its public platform. The company responded with greater emphasis on enterprise software and data licensing.

These companies face different problems. What interests me is how quickly a customer’s reason to pay can weaken once they have another way to do the job.

Software companies face this too, which rather complicates the advice to become one.

Adopting Salesforce can take six to eighteen months, give or take, depending on the company’s size and existing technology infrastructure. I should know. I watched it unfold from the inside in 2020. Still gives me shivers.

So I paid attention when Curative CEO Fred Turner said his company built an internal CRM with AI assistance in two months and canceled a Salesforce contract costing $600,000 a year. A company spokesperson confirmed the cancellation notification. The build time comes from Turner’s account; the long-term performance and costs still deserve scrutiny.

I would hate to be presenting an eighteen-month implementation plan to someone who had just read that.

The comparisons won’t always hold. Large organizations have requirements that a smaller company may never encounter. But customers don’t need to reproduce Salesforce’s entire platform. They need to replace the functions they use, at a cost they can justify.

That gives vendors something more troublesome than another competitor’s price list.

Salesforce’s current results don’t show a mass departure. Quarterly revenue grew 11%, including an acquisition contribution, and reported contract-value attrition remained consistent with the previous year. The company is still growing.

Even so, a customer with a credible alternative has more room to negotiate. A vendor can keep the contract and still lose some of the freedom it once had to set the terms.

Freelancers often have much less room to absorb that pressure.

In Q2 2026, Upwork reported fewer active clients and a 3.6% decline in gross services volume. Its AI-related work grew more than 22%. Fiverr reported annual active buyers down 21.9% and marketplace revenue down 15.5%, explicitly identifying pressure from AI adoption in categories exposed to automation.

There’s money moving into new work. There are also people losing the assignments that kept them afloat.

I find the breezy advice to “move upmarket” particularly hard to stomach. To whom? With what training, paid for out of which disappearing income?

A freelancer can understand the opportunity perfectly well and still struggle to reach it. Calling the transition inevitable doesn’t make it affordable. And telling someone to adapt is remarkably easy when their lost income isn’t coming out of your account.

The pressure reaches the customer, too. Once building your own tools becomes possible, it starts to sound like something you’re negligent for failing to do. Why are you still paying for that software? Why are you hiring someone for that task?

Those can be useful questions. I’d also like to know who has time to act on the answers.

Somebody inside the business has to decide what’s worth building, learn enough to judge whether it works, and stay responsible for it afterward. That work has to fit into a week already occupied by running the business.

I recognize that because we’re doing it.

We run a design studio focused on neuroaccessibility. We spend a lot of time looking at the effort digital products demand from people: what they have to remember, what interrupts them, where they get stuck.

Somewhere in that work, we found ourselves building a suite of apps and platforms for our team and our clients. Individual problems kept giving us a reason to build. An existing tool made a task unnecessarily difficult, or solving it meant adding another subscription with its own demands.

I’m glad we have the ability. I’m also annoyed by how much of it amounts to repairing the experience of using other software.

People already spend hours accommodating their tools. They remember which system holds the latest version, check the dashboard that doesn’t send useful updates, and dismiss notifications that all seem to consider themselves urgent.

For someone whose attention is easily exhausted by that traffic, the effort can eat into the capacity they need for their work. We see a reason to address it. That doesn’t make us eager to inherit an endless support queue.

When our tools break, we own the problem. If someone depends on one to do their job, we can’t lose interest in it because we’ve moved on to something else. The development may have been quick. The commitment lasts as long as people need the thing to work.

That part deserves at least as much attention as how fast we built it.

This is where the advice starts to feel like bullying. A new capability becomes another condition of being taken seriously as a business. If you hesitate, apparently you’re resisting progress. The possibility that you’ve considered the work involved and would rather spend your time elsewhere barely gets a hearing.

Perhaps I like the business I already run. Perhaps I’d prefer to spend the afternoon doing that.

I’m willing to reconsider how we work. I’d like that decision to include the hours we’ll spend maintaining what we build, and the work we’ll have to put aside to do it. Enthusiasm doesn’t remove either from the calendar.

And yes, our own clients are entitled to ask whether they can do more themselves. I can’t defend their freedom to build while expecting them to protect our fees. We have to demonstrate where our involvement improves the work. A speech about the irreplaceable human spirit won’t survive much scrutiny from procurement.

There’s a lot to welcome in having more choices. An expensive contract should be open to challenge. A small team should be able to solve a problem a large vendor keeps ignoring.

I just don’t accept that taking on software development is automatically a happy ending. It can give a business independence. It can also leave its owner with another job they never intended to have, with nobody else available to cover it.

I want to build when the reason is good enough to justify that job. I also want to be able to buy a tool and get back to work.

Apparently, that second option needs defending now.

Sources

Getty Images: October 21 Chapter 11 filing (Reuters)

Getty Images: restructuring announcement, October 12, 2026

Getty Images: generative AI launch, September 2023

Chegg: second-quarter 2026 results

Stack Overflow: CEO interview on Decoder

Curative: Fred Turner on replacing Salesforce (Business Insider)

Salesforce: second-quarter fiscal 2027 results

Salesforce: July 2026 quarterly filing

Upwork: second-quarter 2026 results

Fiverr: second-quarter 2026 results