LIRNEasia Senior Research Manager and Team Lead for Data, Algorithms, and Policy (DAP), Merl Chandana, was interviewed on NewsFirst Sri Lanka’s Face To Face programme on 9 August 2026.
The interview focused on the wonders and dangers of Artificial Intelligence (AI). The discussion covered a recent cybersecurity incident where an experimental agentic AI model developed by OpenAI was manipulated to execute multiple attacks on Hugging Face, an open-source platform and community for AI and machine learning. Agentic AI refers to AI systems that can act autonomously to achieve a goal. Unlike traditional AI systems that mainly respond to user instructions, agentic AI can plan tasks, make decisions, use tools, and take actions with limited human intervention.
During the discussion, Merl highlighted the growing capabilities of agentic AI, the cybersecurity threats it poses, and the challenges associated with governing such systems. He also discussed their potential implications for countries such as Sri Lanka, particularly given the country’s developing digital infrastructure and limited regulatory capacity.
To address the issue in the local context, Merl suggested identifying an appropriate mix of existing laws and policies and considering their application at various levels, including their impact on citizens and the need to update school curricula to build digital literacy. He recommended that, rather than introducing new laws, Sri Lanka should review and amend existing cybersecurity legislation to address emerging risks where applicable.
Merl also explained that while agentic AI has some autonomous capabilities, its agency is not absolute as often portrayed in mainstream narratives such as science fiction. It continues to be shaped by human intent and motivations and remains accountable within the human structures we have built. Citing researchers at Princeton University, Merl urged the public to consider AI as “normal technology” with highly capable functionality that can help humans in predictable ways, similar to other technologies. He noted that attributing undue agency to AI can lead to a misplaced understanding of the issue at hand. Merl highlighted that focusing on AI as an absolute autonomous actor, rather than examining the human decisions and actions behind it, risks addressing the wrong problem.
For those interested, the full Interview can be viewed below.
