AI Faked Taylor Swift's Wedding Photos. The Same Week It Cut 3,230 Jobs.

From Madison Square Garden to the redundancy notice, what AI did in one week.

Taylor Swift and Travis Kelce got married on July 3. They held the ceremony at Madison Square Garden in New York. They had not released any official photos when people started seeing pictures online.

Joseph Kahn directed several of Swift's music videos. He was at the wedding. He posted on X a few weeks later: "Every picture I've seen of the wedding is fake."

That is where we are.

AI Did Not Wait for the Couple to Leave the Building Before Making Fake Versions of Them

The wedding had close to a thousand guests. Adam Sandler officiated. Austin Swift served as Man of Honor and Jason Kelce was best man. The couple wore custom Christian Dior. None of that stopped AI tools from generating convincing images of a wedding that looked almost exactly like the real one, placed in locations where the couple had never been.

An X account posted AI-generated images a few weeks later, claiming to show the newlyweds in Boca Raton, Florida. Some people congratulated them in the comments. Others pointed out that the buildings in the background looked like New York.

The couple had not shared a single photograph. That gap was enough. AI filled it immediately, with images that looked real enough to confuse people who wanted to believe them.

Right of publicity laws protect people from having their name and image used without consent. Those laws vary by state. They do not appear to be moving faster than the tools.

Monday.com Cut 20 Percent of Its Workforce. The CEO Said It Was Not About Cutting Costs.

July 22. Monday.com announced it was cutting 630 jobs, roughly one in five people at the company. Co-founder Eran Zinman put out a statement that same day. He was clear: the decision was not made to reduce costs. It was not made to replace people with AI.

What it was made to do, he said, was restructure the company around a future where AI agents work alongside employees.

Monday.com makes software that helps teams organise tasks, projects, and deadlines. This year it rebuilt its entire product around AI agents, which can now carry out many of those tasks automatically. That shift, the company says, required a different kind of team. So 630 people left.

The timing also matters. Monday.com lifted its margin outlook for 2026 the same week it announced the cuts. The numbers will look better. A fifth of the workforce will not be there to see it.

Visa Cut 2,600 Jobs the Following Week. The CEO Called It Agentic AI.

Six days later. Visa announced it was eliminating 2,600 roles globally, about 7 percent of its total workforce.

CEO Ryan McInerney sent an internal memo. He said AI, specifically what he called agentic AI, was helping to accelerate the evolution and shape the way work gets done at Visa.

The cuts were not just junior roles. A WARN notice filed July 31 with California authorities listed six vice presidents, 37 senior directors, and 16 chief engineering and architect positions at the Foster City campus alone.

The word agentic is worth pausing on. Agentic AI refers to systems that can operate independently, take sequences of steps, and complete tasks without a human approving each one. When a company says agentic AI is reshaping how work gets done, they are describing a system that handles decisions people used to handle. Some of those people are now gone.

3,230 jobs in one week. Two companies. One explanation.

Europe Just Made AI Say What It Is.

On August 2, the EU AI Act transparency rules came into force. They are the first continent-wide rules of their kind.

Chatbots must identify themselves as AI when talking to people. Deepfakes must carry a label. Machine-generated content must carry machine-readable marks. Any AI system analysing emotions or categorising people biometrically must disclose this.

Noncompliance can result in fines of up to fifteen million euros or 3 percent of worldwide annual revenue, whichever is higher.

The debate about whether AI should be required to identify itself is over in Europe. The answer is yes. The US does not have equivalent rules. That gap will matter.

Three things from the past few weeks. AI invented a wedding. AI eliminated jobs. Europe decided AI has to be honest about what it is.

The common thread is that nobody asked permission.