India Proposes AI Labelling After Study Finds 45% Errors

The Indian government has proposed amendments to IT rules that would require clear labelling, traceability and non-removable identifiers for AI-generated content and increase platform accountability to curb deepfakes and misinformation. Experts say the reforms are a necessary step toward responsible AI adoption but will require clear implementation standards and industry–government collaboration. Separately, an international study coordinated by the European Broadcasting Union and led by the BBC, involving 22 public service media organisations and more than 3,000 AI-generated news responses, found that four leading AI assistants (ChatGPT, Microsoft Copilot, Google Gemini and Perplexity) misrepresented news in 45% of answers, with about 31% showing serious sourcing problems and 20% containing accuracy errors. Google’s Gemini performed worst on sourcing (around 72–76% of responses with significant issues), and all models produced basic factual errors, demonstrating systemic cross-border and multilingual failings. Researchers and public broadcasters warn these shortcomings threaten public trust as more people—especially younger users—rely on AI for news and call on tech firms to prioritise error reduction and transparency alongside regulatory measures.