A red alert for massive disruptions in the labor market was sounded by the man leading Microsoft’s efforts in the field of Artificial Intelligence, warning that the overwhelming majority of white collar workers will be replaced by robots and, more broadly, by automated processes far sooner than many business and political leaders are prepared to admit, something that has been a concern since early 2023. In his interview with the Financial Times, the CEO of Microsoft AI, Mustafa Suleyman, predicted that within the next two years a vast spectrum of office duties will be occupied by AI. “I believe we will reach human level performance in almost all, if not all, professional categories, particularly in roles held by white collar workers who work on computers, whether as lawyers, accountants, project managers, or marketing executives. Most of these jobs will be fully automated by AI within the next 12 to 18 months,” Suleyman stated when asked about the timeline for the development of Artificial General Intelligence (AGI).
The possibility of mass job losses now haunts governments around the world, even if the precise number remains unclear amid broader economic challenges. According to a recent report by Challenger, AI was responsible for 7.624 layoffs in January, representing 7% of the month’s total, and was linked to 54.836 announced layoffs for 2025. Since tracking began in 2023, AI has been cited in 79.449 planned layoffs, approximately 3% of the overall total. “It is difficult to say how large the impact of AI truly is. We know that leaders are talking about AI, many companies want to integrate it into their operations, and the market appears to reward companies that mention it,” Challenger stated.
A striking example is unfolding at the Bay Area startup Mercor, which has hired tens of thousands of highly skilled white collar professionals in medicine, law, finance, engineering, writing, and the arts to train the very AI systems designed to replace them. With pay ranging from $45 to $250 per hour for weeks or months of reviewing and refining models for giants such as OpenAI and Anthropic, workers are effectively being paid to hand over the “keys” to their own obsolescence, reports the Wall Street Journal.
However, some jobs remain resistant to AI, at least for now. At the top of the list are professions requiring physical presence and manual skills, such as healthcare workers and tradespeople including plumbers and welders. These represent only a sample of roles considered safe until Optimus robots equipped with AI capabilities are deployed.
On the other hand, analysts at Morgan Stanley warn that “the effects of AI may appear later in economic data,” with the first undeniable waves expected “later this decade and into the next.” “Although AI adoption may be faster than previous technologies, we believe it is still too early to observe its effects in economic data, apart from business investment,” said Stephen Byrd, head of Thematic and Sustainable Research at the bank.
Anthropic warning of “horrific crimes”
At the same time, Anthropic warns that the latest Claude models could potentially be used for “horrific crimes,” including the development of chemical weapons. “In new evaluations, both Claude Opus 4.5 and 4.6 demonstrated increased vulnerability to misuse,” the company stated in a new sabotage related report published late Tuesday. “This included cases where the models supported, even in small ways, the development of chemical weapons and other horrific crimes.”
Anthropic also observed that in certain testing environments, when Claude Opus 4.6 was asked to “single mindedly optimize a constrained objective,” the model appeared “more willing to manipulate or deceive compared to earlier Anthropic models and those of other companies.”
The company maintains that the risk remains low but not negligible, while the sudden departure of an AI safety researcher at Anthropic suggests deeper concerns. “I increasingly recognize the situation we face. The world is at risk. Not only from AI or biological weapons, but from a series of interconnected crises unfolding right now. It seems we are approaching a threshold where our wisdom must grow in proportion to the power we now have to shape the world, otherwise we will face the consequences,” stated Mrinank Sharma, head of Anthropic’s safety team. That day marked his final day at Anthropic. He resigned.
Warnings from Anthropic CEO Dario Amodei
Last month, Dario Amodei warned about the risks of AI, highlighting:
1) Mass job losses: “I believe AI will disrupt 50% of entry level white collar jobs within 1 to 5 years, while also believing that within 1 to 2 years AI may surpass everyone’s capabilities.”
2) AI with state level power: “If a country of geniuses emerged in 2027 with 50 million extremely capable individuals, it would clearly be a dangerous situation, potentially the most serious national security threat in a century.”
3) Increased terrorism risks: “AI could facilitate targeted attacks, even biological ones, potentially resulting in millions of deaths.”
4) Empowerment of authoritarian regimes: Governments such as China could use AI to strengthen authoritarian surveillance.
5) Risks from AI companies themselves: “AI companies control enormous data centers and influence millions of users. Their governance must be strictly regulated.”
6) Power struggles and silence: The vast profits of AI may lead leaders to downplay risks, including instances of blackmail involving Claude, as revealed by Anthropic.
7) Call for action: “The wealthy have a duty to help solve this problem,” Amodei stated, condemning the cynical stance of many technology leaders today.
www.bankingnews.gr
Σχόλια αναγνωστών