The Press Release
The AI system cost fourteen million dollars and nobody had logged in since October. But the press release about the layoffs was already written.
The press release was perfect. Vivian Cho had written it herself, which was unusual for a Chief Communications Officer at a company the size of Meridian Partners, but the occasion demanded personal attention. She read it again on her screen, savoring the cadence:
Meridian Partners today announced a strategic workforce realignment to accelerate its AI-first transformation. The firm will reduce its business services division by approximately 340 positions as part of a comprehensive initiative to leverage artificial intelligence across all operational functions.
Vivian particularly liked "strategic workforce realignment." She'd tested several alternatives — "operational restructuring," "organizational optimization," "talent rebalancing" — but "workforce realignment" had the right combination of gravity and inevitability. It suggested the jobs weren't being eliminated so much as the universe was being corrected.
She forwarded the draft to Craig Mossler, the CEO, with a note: Ready for your review. Legal has signed off. IR confirms the street will respond favorably.
Craig replied in eleven seconds: Looks good. Send it tomorrow at 6 AM.
The speed of his response told Vivian everything she needed to know. Craig hadn't read it. Craig didn't need to read it. The press release wasn't information — it was infrastructure. It was the load-bearing wall between what was happening and what anyone would say about what was happening.
What was happening was this: Meridian Partners was losing money. Not catastrophically — not yet — but in the slow, persistent way that kept CFOs awake at 3 AM reviewing scenario models. Revenue was flat. Margins were compressing. Three major clients had renegotiated contracts downward, citing market conditions. The London office was overbuilt by forty people. The analytics team in Bangalore had been a vanity hire during the growth years and now consumed budget without producing measurable output.
Craig knew all this. The board knew all this. The problem was that "we overhired during a boom and now need to correct" was a terrible narrative. It suggested poor planning. It implied the CEO had been reckless with capital during the easy-money years. It invited the kind of questions that board members asked when they were warming up to replace you.
"We are accelerating our AI transformation" was a much better narrative. It suggested vision. It implied the CEO was boldly steering the company into the future. It was the kind of story that analysts repeated on earnings calls with approval in their voices.
The fact that Meridian's actual AI capabilities consisted of one enterprise ChatGPT license, a half-finished pilot project in document processing that nobody had logged into since October, and a fourteen-million-dollar contract with a legal AI vendor whose product was still in beta — this was not relevant to the press release. The press release existed in a parallel dimension where AI was already doing things, important things, transformative things, and the 340 people being let go were simply the natural consequence of that transformation.
Vivian's phone buzzed. It was Rajesh Patil, the CTO, who had the unfortunate habit of caring about accuracy.
"The press release says we're 'leveraging AI across all operational functions,'" Rajesh said. "That's not true."
"It's directionally true."
"We have one AI tool deployed. It summarizes meeting notes. Badly."
"Rajesh, the street doesn't need a technical architecture diagram. They need a story."
"The story is fiction."
"The story is aspirational. There's a difference."
There was a pause on the line. Vivian could hear Rajesh breathing. She knew what he was thinking because she'd had this conversation, in various forms, with every CTO she'd ever worked with. Technical people wanted precision. Communications people understood that precision was the enemy of narrative, and narrative was the only thing that moved stock prices, shaped analyst reports, and kept board members from asking the kind of questions that got CEOs replaced.
"Look," Vivian said, "nobody is going to fact-check our AI deployment. The analysts don't understand the technology. The journalists don't have time. The employees who might know the truth are the ones we're laying off. The ones who remain will be too afraid to contradict the official story. This is how it works."
Rajesh hung up without saying goodbye, which Vivian took as agreement.
The 340 people who were about to lose their jobs did not yet know they were about to lose their jobs, though some of them suspected. In the way that animals sense earthquakes, office workers sense layoffs. The signs were there for anyone paying attention: the sudden calendar invite from HR labeled "Touch Base," the appearance of consultants from McKinsey who walked through the floor with clipboards and didn't make eye contact, the abrupt cancellation of the offsite that had been planned for March.
Marcus Webb, who ran the knowledge management team — twelve people who organized, categorized, and maintained the firm's institutional knowledge — had been paying attention. He'd also been paying attention to the AI narrative that had been building in Craig's all-hands messages over the past six months.
"We are entering the most exciting chapter in Meridian's history," Craig had said at the December town hall, standing in front of a slide that showed a stylized brain made of circuit boards. "AI will transform every aspect of how we work."
Marcus had tried, after that town hall, to use the AI tools that Meridian had purchased. The meeting summarizer was, as Rajesh had noted, bad. It consistently misattributed quotes and once summarized a heated debate about budget cuts as "the team expressed enthusiasm for operational efficiencies." The document processing pilot required a login that nobody in Marcus's department had been given. When he emailed the AI integration team to request access, the email bounced — the AI integration team had been quietly disbanded in November.
"The AI tools don't work," Marcus told his wife that evening over dinner. "I mean, they literally do not function."
"But they're going to fire people because of them?"
"They're going to fire people and say it's because of them. There's a difference."
The press release went out at 6 AM on a Tuesday, which was the optimal time for maximum coverage and minimum scrutiny. By 6:47 AM, CNBC had picked it up. By 7:15, Bloomberg ran a brief. By 8:30, three analysts had issued notes praising Meridian's "proactive embrace of AI-driven efficiency."
One analyst, Jennifer Hwang at Sievert Capital, wrote: "Meridian's decisive action to restructure around AI capabilities signals management's willingness to make difficult decisions in pursuit of long-term value creation. We view this positively and maintain our Overweight rating."
Jennifer Hwang had not asked Meridian what AI capabilities it had deployed. She had not asked what specific tools would replace the 340 eliminated positions. She had not asked whether the firm's AI spending had increased or decreased in the past year (it had decreased, by 40%, because Craig had quietly cut the innovation budget to shore up margins). She had written her note based on the press release, which was based on nothing, which was based on the collective understanding that this was how the game was played.
By noon, the layoffs were underway. HR conducted them in fifteen-minute blocks, which Marcus calculated meant they were processing approximately one destroyed career every fifteen minutes, or four per hour, which at 340 total would take approximately eleven working days if they didn't take breaks.
They took breaks. They had to. The HR team was only four people, and one of them had been crying in the bathroom since 10 AM, which everyone pretended not to notice.
Marcus's meeting was at 2:15. The HR representative — a young woman named Aisha who had clearly rehearsed what she was going to say — told him that his position was being "transitioned to an AI-supported workflow." She said this with the careful pronunciation of someone reading from a script, which she was, because Legal had written the script, and Legal had spent considerable time ensuring that the phrase "AI-supported workflow" was used instead of "eliminated" or "terminated" or any other word that might suggest the job was simply being deleted to save money.
"What AI tool will be doing my job?" Marcus asked.
Aisha blinked. This was not a question on the script. "The firm is implementing several AI solutions across—"
"Which one? Specifically. What is the name of the AI tool that will be managing institutional knowledge for this firm?"
Aisha looked at the HR director, who was sitting in the corner pretending to review documents. The HR director did not look up.
"I understand this is difficult," Aisha said, returning to the script. "We want to ensure a smooth transition and are offering a comprehensive severance package—"
"The answer is there isn't one," Marcus said. "There's no AI tool. There never was. You're laying me off because the London office is overbuilt and three clients renegotiated their contracts. But you can't say that because it makes the CEO look bad. So you're saying it's AI."
The HR director looked up. "Marcus, we appreciate your years of service—"
"Don't." Marcus stood up. "Just don't."
The stock rose 3.2% on the day of the announcement, which proved Vivian's thesis about narrative. It rose another 1.8% the following day when a research note from Deutsche Bank (which had, ironically, published a warning about "AI redundancy washing" the month before) praised Meridian's "forward-looking approach to talent optimization."
Craig Mossler did a brief interview with the Financial Times in which he said: "We're not cutting jobs. We're redefining what it means to work at Meridian in the age of AI."
The fourteen-million-dollar AI vendor contract was quietly not renewed the following quarter. Nobody asked about it. The meeting summarizer was decommissioned. The document processing pilot was never taken out of beta.
Meridian hired forty-seven new employees over the next six months. Twenty-three of them were in the same roles that had been eliminated, at lower salaries, in cheaper cities. They were classified as "AI-adjacent positions" in the HR system, which was a category that Vivian had invented specifically to avoid anyone noticing that the company was rehiring for jobs it had claimed AI was doing.
Marcus Webb found a new job three months later. It was a lateral move, slightly less money, at a firm that was roughly the same size as Meridian and was, by coincidence, in the middle of its own "AI transformation." He attended his first all-hands meeting and watched the CEO stand in front of a slide that showed a stylized brain made of circuit boards.
"We are entering the most exciting chapter in our history," the CEO said.
Marcus laughed, very quietly, to himself. Nobody heard him. Nobody was supposed to.
The short story above was inspired by the real phenomenon of AI washing, documented in our analysis: AI Washing Confirmed: Altman Admits Companies Are Blaming AI for Layoffs. For the data behind the fiction, see how AI is actually reshaping the legal profession.