BIS Warns: AI Is Complicating Inflation Assessment for Central Banks

The Bank for International Settlements, an institution often described as the central bank for central banks, warns in its latest annual report that the rapid rise of artificial intelligence is changing how inflation must be measured and how interest rates must be set. According to the BIS, policymakers can no longer treat AI as a distant technological trend, but must recognize it as a structural shift comparable to electrification or the spread of the internet.

Four Flashpoints

In its annual economic report, published in late June, the BIS identifies four main sources of risk to the global economy in 2026: renewed inflationary pressure following the energy shock tied to the Middle East conflict, uncertainty over the durability of the AI investment boom, growing vulnerabilities in the non-bank sector, and weakening public finances in advanced economies. The institution stresses that the global protective buffers that helped absorb earlier shocks are visibly shrinking.

The closure of the Strait of Hormuz and the conflict in Iran, the BIS calculates, wiped out more than 10 million barrels of oil per day, or about 13 percent of normal supply. That translated into roughly a 0.5 percentage point rise in global inflation. The bank warns that after the experience of 2021-2023, central banks cannot dismiss the risk of second-round effects, in which a one-off price increase turns into more persistent inflationary pressure and inflation expectations become unanchored.

AI as a New Macroeconomic Variable

The biggest novelty in this year's report is the place the BIS gives artificial intelligence in its inflation analysis. The debate has shifted from whether AI matters for the economy to questions of timing, direction and transmission mechanism: how quickly the effects will appear, whether they will first push prices up or down, and how central banks should respond to a force that may be inflationary in the short term and deflationary in the long term.

On one hand, massive spending on semiconductors, data centers and energy infrastructure is boosting global growth and financial markets, but at the same time the electricity hunger of this buildout is already pushing up prices and production costs, with possible knock-on effects for inflation. On the other hand, the BIS acknowledges that AI could turn out to be a disinflationary force if the promised productivity gains actually materialize in the coming years.

The rapid and widespread adoption of artificial intelligence means there is an urgent need for central banks to upgrade their skills - from the Bank for International Settlements' annual report

How Central Banks Already Use AI

Some institutions are not waiting for this debate to be settled and are already turning to artificial intelligence to analyze their own policy. The European Central Bank has used a machine learning model analyzing around 60 indicators since late 2022, which in the second and fourth quarters of 2025 signaled a rise in core inflation roughly 20 basis points ahead of the actual reading.

Germany's Bundesbank confirmed in December 2025, through its president Joachim Nagel, that it uses a wide range of AI applications, including a model called MILA, which analyzes communications from euro area central banks. At the Federal Reserve, Christopher Waller and Philip Jefferson have publicly discussed the productivity effects generated by AI, while another FOMC member, Kevin Warsh, has warned of a phenomenon described as the "escape velocity" of inflation driven by AI investment.

Debt, Stablecoins and Overinvestment Risk

The BIS also points to the other side of the AI boom, namely how it is being financed. The scale of investment commitments by major technology companies is starting to outpace their current earnings, and intense competition among hyperscalers may push them to commit capital to projects with uncertain returns. The bank warns of the risk of "circular financing", meaning companies within the same AI ecosystem financing one another, which could trigger a domino effect if demand were to collapse.

The report also highlights the growing role of stablecoins in financing this infrastructure buildout. Their market capitalization reached around $320 billion by the end of May 2026, with 99.4 percent backed by dollar-pegged fiat currencies. The estimated annual transaction volume in stablecoins reached $28 trillion in 2025.

What It Means for Policymakers

For central banks, the practical takeaway from the report is that it is no longer enough to watch AI's development from the sidelines. The BIS stresses that monetary and supervisory institutions must themselves become users of the technology in order to effectively assess and manage the risks associated with it, rather than remaining, as the report puts it, merely informed observers of technological progress.

This translates directly into the job market inside central banks themselves, where demand is growing for analysts who can combine economic expertise with machine learning skills. The gap between banks that have already deployed tools such as the ECB's model or the Bundesbank's MILA, and those still building such capabilities, could in practice mean differences in the accuracy of inflation forecasts and the speed of response to economic shocks.

For Polish companies and investors, the BIS's message boils down to this: decisions by the Federal Reserve, the European Central Bank and other key institutions in the coming quarters will increasingly have to account for unusual, hard-to-forecast factors tied to AI investment, such as energy prices around data centers or wage growth in AI-related occupations. That adds to uncertainty over the path of interest rates, which drives borrowing costs and asset valuations well beyond the United States and the euro area.