Banks rate AI-exposed firms safer but narrow leaders' advantage
Bruno Buchetti, Christian Eufinger, Luca X. Lin and Ixart Miquel-Flores find banks assign lower default risk to AI-exposed firms but narrow market leaders' relative advantage. European Central Bank Working Paper 3302 was published October 19, 2026.
Safer industries, less protected leaders
A one-standard-deviation rise in AI attention is associated with roughly 5 percent lower assessed default probabilities for borrowers in the most exposed industry quartile relative to others.
Within highly exposed markets, leaders' assessed risk rises about 8 percent relative to the smallest competitors; leaders do not become riskier in absolute terms.
Their relative drawn and committed credit fall 3.8 and 2.4 percent.
Relationship exits rise 1.2 percentage points from a 7.1 percent base, while new-lender entry falls 0.6 points from 2.2 percent.
Leaders' advantage in avoiding bad-credit events over the next year narrows by 0.5–0.7 percentage points against an 8.2 percent event rate, even controlling for lenders' consensus default assessments.
Sector knowledge outweighs borrower familiarity
The European Central Bank (ECB) paper matches AnaCredit and Orbis data for private non-financial firms in Germany, France, Italy and Spain over Q1 2021–Q4 2024: roughly 6.8 million firm–bank–quarter observations.
Five industry exposure measures meet Google and Wikipedia attention indices around ChatGPT's Q4 2022 release.
Sector specialists lower assessed risk 11–16 percent more than co-lenders per standard deviation of attention; long relationships add no differential.
Risk falls about 7 percent for less labor-intensive firms versus 3 percent for more labor-intensive firms.
Beliefs are not realized gains
The results measure banks' beliefs, not proven AI productivity gains.
Industry exposure does not establish firm adoption, and credit changes do not separate supply from demand.
The authors' views are not those of the ECB; specialist optimism remains unvalidated.