Global regulators have stated that increasingly autonomous forms of artificial intelligence could heighten risks to the financial system and have called for new safeguards as its adoption accelerates.
The Financial Stability Board (FSB), in a report published Wednesday, urged boards of directors to consider implementing safeguards to mitigate risks associated with artificial intelligence, including “agent-based” AI—or systems capable of planning, reasoning, and performing tasks with limited human involvement, Reuters reports.
The FSB, a global standards-setting body, stated that autonomous AI poses risks that could “materialize rapidly,” including the potential for unauthorized or illegal actions, data breaches, and disruptions to connected systems.
“Artificial intelligence poses a serious challenge to human control,” the report states, warning that agents may perform actions that deviate from companies’ intentions without staff knowledge or the ability to intervene quickly.
To address these issues, the standards body presented a series of proposed “best practices,” urging financial firms to establish clear boundaries for AI use and implement safeguards. These non-binding recommendations are open for feedback until July 22.
AI in Finance
Financial firms are already using agent-based AI for fraud detection, customer service, and internal operational functions. They also impose restrictions on the actions of AI agents and require human approval for high-risk actions, such as financial transactions exceeding certain thresholds.
Banks have stopped implementing neural networks just to follow the trend and now require them to deliver clear financial returns. Often, the costs of AI integration do not justify themselves, leading to financial distortions and erroneous forecasts.
Nearly 52% of respondents from the financial sector who participated in the Cambridge Centre for Alternative Finance survey reported actively implementing agent-based models, with 23% scaling or transforming them and 29% conducting pilot projects to implement agent functions.
Companies may also consider adapting human resources management mechanisms and processes to AI agents so that they are treated as “synthetic employees,” the FSB stated.
