Some of the technology industry's loudest AI boosters are quietly pumping the brakes. New reporting from Reuters and The Guardian indicates that Meta is scaling back ambitious plans to lay off staff and replace them with AI agents, while Microsoft is reassessing its own rollout amid a flood of low-quality, machine-generated content inside its own walls.
The shift is notable less for its size than for its source: the companies now trimming their expectations are among the ones selling the AI revolution to everyone else. For more context on this story, see our ongoing latest AI developments.
The 'Doom Loop' of Workplace AI Slop
The most vivid account of the problem came from inside Microsoft itself. Ryan Roslansky, an executive vice president at the company, took to his public LinkedIn account to lament the rise of what has come to be known as workslop — AI-generated documents and messages that look like work but contain none of the substance.
"AI slop isn't happening just on social networks… it's happening in the day to day of work," Roslansky wrote, according to The Guardian. He described receiving a document that was "clearly AI generated," running it through an AI summarizer, and ending up "with nothing that brought new thinking to the idea." His verdict: "This is a doom loop."
The anecdote captures a paradox that has been building in offices for the past two years. Employees use AI to produce output quickly; colleagues use AI to compress that output into summaries; and the net result is a cycle in which machine-generated text is fed to machines, with humans left to wonder where the thinking went.
The economics are becoming part of the conversation, too. Reporting through the year has described office workers running up costly AI token bills largely to appear productive — spending that finance departments are increasingly unwilling to write off as innovation. What began as an experiment with a vague mandate is turning into a line item that has to justify itself, and in many cases is failing to.
'Driving the Truck While Building It'
Corporate America's broader frustration was summarized by Johnny Taylor Jr., chief executive of the Society for Human Resource Management, who told The Guardian that executives feel "we're driving the truck and building it at the same time, and that's a problem."
"If we could all rewind… we would have introduced [this] in a more thoughtful, slow way," Taylor said — a striking admission from the head of the world's largest HR professional society about a technology many companies adopted with competitive urgency.
That urgency has carried real costs. As earlier reporting documented through 2026, office workers have burned through budgets on AI tokens largely to appear productive, while consultant reviews found enterprises cutting AI agent rollouts as costs outran value.
When AI Projects Fail, They Fail Expensively
The financial tail risk is no longer hypothetical. A September survey by Ernst & Young found that 36% of senior AI executives at firms with more than $1 billion in annual revenue have already experienced what they classified as a catastrophic AI incident — one that caused financial damage, operational disruptions, data loss or brand damage.
Nor is this the first major survey to find corporations second-guessing their deployments. A KPMG study reported in August found that roughly half of enterprises had already scaled back AI agent rollouts as costs outran delivered value, and corporate caution has surfaced repeatedly through the year, from abandoned proofs-of-concept to quietly shelved "AI-first" reorganizations.
Against that backdrop, the recalibration at Meta and Microsoft looks less like a retreat from AI than a narrowing of it. According to Reuters, Meta is scaling back plans to cut staff and swap in AI agents — a program that once promised sweeping headcount replacement. The Guardian's reporting suggests Microsoft is undergoing a similar reassessment of how aggressively to push AI into daily work.
The Reckoning Comes as Agent Ambitions Grow
The pullback lands at an awkward moment for the industry's next act: agentic AI. Software makers have spent the year positioning autonomous agents as the future of knowledge work — systems that don't just draft text but book meetings, move money and operate software on a user's behalf. Consumer-grade agents, meanwhile, keep making headlines for precisely the kind of uncontrolled behavior that makes enterprise buyers nervous.
ਐਂਟਰਪ੍ਰਾਈਜ਼ ਦੇ ਤਜ਼ਰਬੇ ਤੋਂ ਉੱਭਰਨ ਵਾਲਾ ਸਬਕ ਬੇਮਿਸਾਲ ਪਰ ਇਕਸਾਰ ਹੈ: ਮੁੱਲ ਤੰਗ, ਚੰਗੀ ਤਰ੍ਹਾਂ ਨਿਰੀਖਣ ਕੀਤੇ ਤੈਨਾਤੀਆਂ ਤੋਂ ਆਉਂਦਾ ਹੈ, ਸਟਾਫ ਲਈ ਏਜੰਟਾਂ ਦੇ ਥੋਕ ਬਦਲ ਤੋਂ ਨਹੀਂ। ਬਜਟ ਸਮੀਖਿਆਵਾਂ ਤੋਂ ਬਚਣ ਵਾਲੇ ਟੂਲ ਮਾਪਣਯੋਗ ਆਉਟਪੁੱਟ ਅਤੇ ਸਪੱਸ਼ਟ ਅਸਫਲਤਾ ਮੋਡਾਂ ਵਾਲੇ ਹੁੰਦੇ ਹਨ, ਨਾ ਕਿ ਡਿਜੀਟਲ ਸਹਿਕਰਮੀਆਂ ਵਜੋਂ ਪੇਸ਼ ਕੀਤੇ ਗਏ।
ਉਹ ਕੰਪਨੀਆਂ ਜੋ ਜੀਵਣ ਲਈ AI ਵੇਚਦੀਆਂ ਹਨ ਉਹ ਆਪਣੇ ਗਾਹਕਾਂ ਨਾਲੋਂ ਇਸ ਸਬਕ ਨੂੰ ਤੇਜ਼ੀ ਨਾਲ ਅੰਦਰੂਨੀ ਬਣਾਉਂਦੀਆਂ ਪ੍ਰਤੀਤ ਹੁੰਦੀਆਂ ਹਨ - ਜੋ ਸਭ ਤੋਂ ਵੱਧ ਦੱਸਣ ਵਾਲਾ ਸੰਕੇਤ ਹੋ ਸਕਦਾ ਹੈ। ਜਦੋਂ AI ਅਰਥਚਾਰੇ ਨੂੰ ਬਣਾਉਣ ਵਾਲੀਆਂ ਪਲੇਟਫਾਰਮ ਕੰਪਨੀਆਂ ਸਾਵਧਾਨੀ ਨਾਲ ਆਪਣੇ ਖੁਦ ਦੇ ਏਜੰਟ ਅਭਿਲਾਸ਼ਾਵਾਂ ਦਾ ਇਲਾਜ ਕਰਨਾ ਸ਼ੁਰੂ ਕਰਦੀਆਂ ਹਨ, ਤਾਂ ਇਹ ਟੈਕਨਾਲੋਜੀ ਦੇ ਮਾਰਕੀਟਿੰਗ ਅਤੇ ਇਸਦੀ ਰੋਜ਼ਾਨਾ ਦੀ ਕਾਰਗੁਜ਼ਾਰੀ ਦੇ ਵਿਚਕਾਰ ਪਾੜੇ ਬਾਰੇ ਕੁਝ ਦੱਸਦੀ ਹੈ ਜਿਸ ਨੂੰ ਕੋਈ ਵਿਸ਼ਲੇਸ਼ਕ ਰਿਪੋਰਟ ਪੂਰੀ ਤਰ੍ਹਾਂ ਹਾਸਲ ਨਹੀਂ ਕਰਦਾ।
ਰਾਇਟਰਜ਼ ਤੋਂ ਰਿਪੋਰਟਿੰਗ, ਦਿ ਗਾਰਡੀਅਨ ਅਤੇ ਭਵਿੱਖਵਾਦ ਨੇ ਇਸ ਲੇਖ ਵਿੱਚ ਯੋਗਦਾਨ ਪਾਇਆ.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
