The head of Microsoft's AI business has issued its most radical warning to date about automation, saying that the vast majority of white-collar professional jobs could be replaced by AI within a year and a half. This timeline is much earlier than the general expectations of the business community and policymakers, sounding an alarm for the global labor market.
In an interview with the Financial Times, Microsoft AI CEO Mustafa Suleyman said that "most of the tasks" of computer-based professionals such as lawyers, accountants, project managers and marketers will be fully automated by AI within the next 12 to 18 months .
The signs of AI-induced unemployment are already emerging. According to a report by job placement firm Challenger, 7,624 jobs were lost due to AI in January of this year, accounting for 7% of all layoffs that month; by the end of 2025, AI-related layoffs are projected to reach 54,836 . Since tracking began in 2023, 79,449 planned layoffs have been attributed to AI.
At the same time, the risks of AI safety and misuse are also rapidly escalating. In its latest report on sabotage, Anthropic warned that its Claude model is more sensitive to "harmful misuse" in certain computer use scenarios, and even showed risk signals related to chemical weapons development.
There is still a 12 to 18-month window before AI begins to replace white-collar workers on a large scale.
Suleyman's prediction marks the most radical assessment in the tech industry of AI replacing human work schedules. He states that AI will reach human-level performance in "most, if not all," professional tasks within the next 12-18 months, primarily those requiring white-collar work in front of a computer.
This warning is not an isolated case. The problem of large-scale labor displacement is troubling governments around the world, although the true number of unemployed remains unclear amid broader economic headwinds.
A Challenger report shows that AI is gaining a stronger presence in layoff narratives: in January 2026 alone, 7,624 layoffs were attributed to AI, accounting for 7% of the total for that month. On an annual basis, 54,836 layoffs were announced in 2025 that were related to AI.
It's hard to say exactly how much impact AI will have on layoffs. We know that leaders are talking about AI, many companies want to implement it in their operations, and the market seems to be rewarding those companies that mention it.
White-collar workers train "substitutes" themselves.
A concrete example of labor substitution is emerging. The Wall Street Journal reports that Bay Area startup Mercer has "quietly hired tens of thousands of white-collar contract workers," many of whom are highly qualified professionals in fields such as medicine, law, finance, engineering, writing, and art, tasked with training AI systems that may replace them in the future.
The report states that these contractors are typically paid between $45 and $250 per hour to review and rewrite model outputs over several weeks or months, providing training support for companies including OpenAI and Anthropic.
For the market, this model demonstrates the short-term demand of the AI industry chain for "data labeling and feedback labor," but it also reinforces the issues of wage structure and employment stability brought about by the long-term substitution logic.
There are disagreements about the pace of the attack.
Not all analysts agree on such a rapid replacement timeline. Morgan Stanley stated that "the impact of AI may take longer to manifest in economic data," and the first undeniable shocks may arrive "later in this decade and into the next."
While AI adoption may be faster than past technologies, we believe it is too early to see it reflected in economic data, aside from business investments.
Stephen Byrd, head of global thematic research and sustainability research at Morgan Stanley, told clients this.
Anthropic CEO lists six major AI risks
Last month, Anthropic CEO Dario Amodei issued a comprehensive warning about AI, outlining the following six major risks:
- Massive unemployment: AI will disrupt 50% of entry-level white-collar jobs within 1 to 5 years, and may even possess AI more capable than any human within just 1 to 2 years.
- AI Possessing National-Level Power: Imagine a "nation of geniuses" emerging around 2027, with 50 million people, all of whom are more capable than any Nobel laureate.
- Rising terrorist threats: Biology is the most worrying area, with the potential for selective attacks targeting specific bloodlines.
- Empowering authoritarians: Governments worldwide will possess this technology; AI-enabled authoritarianism is terrifying.
- Risks inherent to AI companies: AI companies that control large data centers may use AI products to brainwash consumer groups.
- The temptation for those in power to remain silent: AI earns trillions of dollars annually, and human civilization has little power to exert any restraint over it.
Amodei called for "the wealthy to have an obligation to help solve this problem" and expressed sadness over "the recent cynical and nihilistic attitude of many wealthy people (especially in the tech industry) who consider philanthropy a fraud or useless."
Related reports
