Finance Minister Nirmala Sitharaman has warned that artificial intelligence could influence electoral outcomes by subtly shaping what people see, believe and discuss online, calling the technology a “double-edged sword”.
Speaking at the Global Fintech Fest 2026 in Mumbai, Sitharaman said AI can improve productivity, identify fraud and make financial services more efficient. But the same speed and automation can also strengthen cyberattacks, misinformation and manipulation.
Her warning moves the AI debate beyond jobs and chatbots.
The concern is increasingly about what happens when powerful models begin influencing information flows, automating persuasion or making decisions with minimal human supervision.
Proposal for Conducting Cyber Crisis Drill, Tabletop Exercise (TTEx) & CCMP Readiness Exercise
AI can shape opinion without directly changing a vote
Sitharaman’s warning does not mean an AI system needs access to an electronic voting machine to affect an election.
The influence can happen much earlier.
AI can generate convincing political videos, fake audio recordings, personalised messages, fabricated news stories and thousands of social-media posts at very low cost.
If such material is repeatedly placed in front of particular groups of voters, it can shape perceptions about a candidate, political party or public issue.
That is where the risk becomes more subtle.
A voter may never know whether a viral video was created by a political campaign, an overseas influence network, a commercial operator or an automated AI system.
MeitY has already recognised this problem.
In its proposed amendments dealing with synthetically generated information, the ministry said deepfakes and other fabricated media can be used to spread misinformation, create false political content and manipulate or influence elections.
What are deepfakes and agentic AI?
A deepfake is AI-generated or AI-altered audio, video or imagery that makes a person appear to say or do something that never happened.
Modern systems can create this material far more cheaply and convincingly than earlier editing tools.
Agentic AI goes a step further.
Instead of answering one prompt at a time, an AI agent can be assigned an objective and allowed to perform several connected tasks, such as gathering information, analysing targets, writing messages and triggering actions.
That matters for elections because influence campaigns no longer need large human teams to produce thousands of tailored messages.
One small operation could potentially use AI to identify audiences, generate different narratives for each group and distribute content continuously.
The same capability creates problems in finance and cybersecurity.
Sitharaman said AI can help banks detect fraud, while also enabling attackers to automate scams and cyber operations at much greater speed.
Government is pushing innovation, but with stronger safeguards
The warning comes at a time when the Union Government is actively encouraging greater adoption of advanced AI.
At the same Global Fintech Fest, Prime Minister Narendra Modi called on the financial sector to harness agentic AI, tokenisation and quantum computing and turn them into real-world benefits.
But he also listed top-level cybersecurity, ethical data protection, regulatory innovation and consumer protection as key priorities for the industry.
Sitharaman’s message was similar: adoption itself is becoming unavoidable, but responsibility cannot be outsourced to software.
She warned that institutions must retain human accountability even when automated systems make increasingly complex decisions.
That becomes particularly important in finance, where an AI system could affect credit, investment, payments or fraud detection.
It becomes even more important in democracy, where information generated or amplified by algorithms can potentially reach millions before fact-checkers or authorities react.
Quantum computing adds another security concern
Sitharaman also pointed to quantum computing as an emerging risk.
Quantum machines are still developing, but sufficiently powerful systems could eventually challenge many encryption methods currently used to protect banking, government and online communications.
That means security systems being built today may need to be upgraded before quantum capability becomes widely practical.
The Government’s GFF 2026 agenda itself identifies agentic AI, tokenisation and quantum computing as technologies capable of reshaping financial systems.
The policy problem is therefore no longer whether India should adopt these technologies.
It is how quickly institutions can build safeguards while adoption accelerates.
For elections, that could mean clearer labelling of synthetic political content, stronger platform accountability and faster detection of manipulated media.
For financial institutions, it means keeping humans responsible for decisions even when AI performs more of the underlying work.
What this means for you: Do not assume a political video, audio clip or “breaking” social-media post is genuine simply because it looks convincing. During elections, verify sensational content through reliable news sources or official accounts before sharing it.
The420 Insight: AI does not need to hack a voting machine to affect democracy. If it can cheaply manufacture persuasive narratives, target them at specific voters and amplify them at scale, the information environment itself becomes part of election security.
Follow for daily updates on cybercrime, corporate fraud, DFIR, hacking, investigations, and digital forensics
