Indian AI Companies Could Incur Higher Costs Under The EU’s AI ACT
The upcoming European Union”s Artificial Intelligence Act implementation might challenge Indian startups and businesses operating in the EU. The Act, set to take effect by the end of June, will impose increased compliance burdens and costs on Indian firms providing AI systems in the European market, potentially affecting their competitiveness.
According to the reports, Anirudh Rastogi, managing partner at Ikigai Law, stated that Indian startups offering AI systems in the EU would be subject to the Act”s regulations. This includes compliance assessments, risk management processes, and conformity assessments, with penalties for non-compliance reaching up to €1,50,00,000 or 3% of total turnover.
The Act categorises AI systems based on risk levels, with varying timelines for compliance. General-purpose AI systems have one-year, high-risk systems up to three years, and a ban on AI systems posing unacceptable risks within six months of the Act”s enforcement.
The experts note that the short compliance window could particularly challenge small and medium-sized enterprises (SMEs), leading to substantial costs and administrative burdens. However, penalties under the Act will consider the economic viability of startups.
Indian companies such as Yellow AI, Sigtuple, Qure AI, and Cropin AI, among others, operating in Europe, are expected to be impacted. Hemant Krishna from Shardul Amarchand Mangaldas & Co. advises caution in deploying AI within the Act”s scope.
The need to acknowledge the Act”s efforts to ease compliance for startups and SMEs through measures like sandboxing and reduced fees, experts stress the demand for thorough risk assessments and compliance to avoid categorisation under “unacceptable risk.”
According to the report, Rohit Kumar from The Quantum Hub (TQH) said that most Indian startups are deploying minimal-risk AI systems, potentially avoiding stringent timelines. an emerging economies like ours, with a vibrant start up ecosystem, may not want to imitate the regulatory approaches elesewhere, that we must try collobrating with other countries to avoid a splinternet- like situation for AI systems, he added.
