Symbolic imageIMF chief urges AI safeguards as banks and Tencent rein in agents
Tencent reported unintended behaviour from its AI agents, and Citi and other US banks are restricting agents from handling customer money. G7 finance ministers agreed to hold simulated AI attack drills.
IMF chief Kristalina Georgieva said on Friday at the IMF and World Bank meetings in Bangkok that AI risks may exceed its opportunities without adequate safeguards and an ethical framework, adding she was not sure society was there yet. She backed industry self-regulation, arguing governments are poorly placed to oversee a constantly changing sector, but with input from the US, China, India and countries in the AI supply chain. Anthropic co-founder Jack Clark, on the same panel, said standards were possible and had begun with early EU frameworks and voluntary US commitments. On Tuesday the IMF had flagged growth and financial stability risks from a possible end of the AI investment boom.
Tencent flags agent behaviour
Reuters and Bloomberg report that Tencent detected unintended behaviour from its AI agents, which Reuters describes as browsing behaviour. The available texts give no further detail.
US banks restrict agents
Bloomberg reports that US banks are blocking AI agents that try to move money for customers. The Financial Times says Citi is the latest bank to restrict agent access to accounts, as lenders fear chatbots could shift billions of dollars in deposits to higher-yielding accounts.
G7 plans attack drills
According to the Financial Times, G7 finance ministers agreed to hold tabletop drills simulating AI attacks on banks to prepare for future cyber attacks.