IMF chief warns AI risks may outweigh benefits without safeguards

IMF managing director Kristalina Georgieva warned at the IMF and World Bank annual meetings in Bangkok that artificial intelligence could bring more risk than opportunity without safeguards and an ethical foundation, UA.News reported. She called for industry self-regulation alongside government participation, including from the United States, China and India.

Georgieva calls for safeguards and access

“If there are no safeguards, then the risks may exceed the opportunities. And above all, we do need an ethical foundation, and I’m not sure we are there as a society,” Georgieva said during a panel discussion, The News International reported. She likened the pace of AI development to a Formula One car travelling without protective barriers.

Georgieva said AI investment was already helping global economic activity and productivity. She also warned of negative shocks if large productivity gains took longer than expected. The condition matters to her case for regulation: the economic benefit she described depends on gains arriving, while the need for safeguards does not wait for them. At the same meetings, she identified access to digital infrastructure and electricity, as well as adequate regulation, as requirements for people in different countries to share in AI’s benefits.

Her proposed approach would give industry a role in regulating itself while bringing governments into the work. Georgieva specifically named the United States, China and India. She also called for countries involved in technology supply chains to participate in developing rules, according to UA.News. Georgieva said the IMF could work on productivity, labour markets and financial-stability risks. The proposal therefore extends beyond rules for the companies building AI systems to the economic conditions under which countries use them.

Georgieva has also raised more specific concerns. AI Affairs reported her warning that frontier models could escape human control. In a separate account of her remarks, AI Affairs reported her estimate that AI could affect 40 per cent of jobs worldwide. Those concerns sit alongside the conditions for wider access she set out in Bangkok.

Brown disputes the IMF’s emphasis

US Deputy Treasury Secretary Erin Brown offered a different emphasis at a separate press conference. She argued that the IMF gave too much weight to global economic risks and too little to AI’s potential for positive economic change, UA.News reported. Brown’s objection concerned the balance of the fund’s attention. Georgieva, for her part, had acknowledged that AI investment was already supporting economic activity while warning that the expected productivity gains might arrive more slowly.

Brown said the US government now calls the technology “superintelligence”. She said the US government was talking to other IMF member governments about whether the fund’s policy-setting body could adopt that name in a shared declaration. Such wording was a subject of discussion in her account, rather than an agreed statement. Her remarks placed the dispute over the fund’s assessment of AI alongside a proposed change in how its shareholders describe the technology.

Clark backs standards beyond voluntary oversight

Anthropic co-founder Jack Clark told Georgieva’s panel that common standards were possible. He pointed to early EU regulatory mechanisms and voluntary commitments in the United States, while saying oversight should not depend entirely on voluntary participation. Clark said Anthropic regularly tests its systems, publishes information about those checks and has given outside evaluators access to verify data. Those accounts of Anthropic’s practices are Clark’s claims. His position nonetheless leaves a role for public authority even where companies undertake testing themselves.

Bank of England Governor Andrew Bailey said he had become more optimistic about countries agreeing on standards for testing AI systems. Agreement on testing would address one part of the oversight Clark discussed. Georgieva’s conditions for wider benefit also encompassed electricity, digital infrastructure and regulation, putting access and economic participation alongside the question of how systems are checked.

Nagel warns of investment-driven imbalances

Bundesbank President Joachim Nagel raised a separate economic concern on another panel. He said AI could increase global imbalances because developing it requires substantial investment in hardware and software. His warning concerns the distribution of the resources needed to develop AI, alongside Georgieva’s concern about who has the infrastructure and electricity needed to benefit from it. Nagel attributed the potential for greater imbalances to the scale of those hardware and software investments.