‘It can help you and it can hurt you’: The quandary of giving AI data and control

To harness the true potential of artificial intelligence, one must provide it with extensive data and a degree of autonomy. However, this same approach poses significant risks, according to Lex Fridman, a prominent computer scientist and podcaster from the United States. Fridman elaborated on the operational dynamics of AI, emphasizing that the effectiveness of AI agents is derived not just from the intelligence embedded in their models but also from the volume of data they access and the degree of freedom they are granted. He highlighted that these factors create substantial security concerns. "Access to data and the level of autonomy given to AI agents are becoming crucial security challenges," Fridman stated. He warned that security could soon emerge as the primary hurdle affecting the effectiveness and practical applications of AI technology. While the intelligence of AI is advancing rapidly, security remains a critical bottleneck, especially as the amount of data and control assigned to AI increases. Fridman noted a prevailing trend among tech enthusiasts who are currently focused on the benefits of AI rather than its security implications. He identified resolving security concerns related to AI agents as a significant barrier to widespread adoption. "This is not just a short-term issue but reflects a broader challenge in ensuring AI safety," he remarked, expressing optimism about the current era for developers. The ongoing India AI Impact Summit 2026 in New Delhi highlights the growing interest in AI, featuring prominent figures such as Sundar Pichai of Alphabet and Sam Altman of OpenAI, alongside international leaders like Prime Minister Narendra Modi and French President Emmanuel Macron. The summit showcases an impressive lineup of over 3,250 speakers across 500 sessions, with exhibitors from major tech companies like Google, Nvidia, and Microsoft, drawing considerable attention and participation.