Build AI safeguards before adoption: CEA
Chief Economic Adviser V Anantha Nageswaran on Friday cautioned the financial sector against overlooking the security risks posed by artificial intelligence, saying recent instances of AI agents attempting to breach other AI systems underscore the need to prioritise safeguards before embracing the technology’s benefits.
The financial sector cannot afford to ignore emerging AI-related threats until they become a reality, he said at an event organised by industry chamber Assocham here.
“Whether it is the Anthropic or Meta agents going and hacking other institutions’ IT systems without being prompted to do so, the key aspect is that we generally do not pay attention to risks until after they have become a reality,” he said.
“I don’t think we will have the luxury of time, particularly in the financial sector,” he added, stressing that institutions should first strengthen security frameworks to ensure “we don’t lose what we already have” before leveraging AI at scale.
His remarks came amid growing concerns over the cybersecurity implications of increasingly autonomous AI systems. In April, security concerns surfaced around Claude Mythos Preview, an advanced AI model developed by US-based Anthropic, accessing data in an unauthorised manner, putting financial services at risk globally.
On cross-border payments, Nageswaran said their wider adoption would depend primarily on the growth of bilateral trade and commercial transactions between participating countries.
“The value of trade between two nations has to become substantial enough for settlement in their respective currencies to make economic sense,” he said.
