Alphabet CEO Sundar Pichai has warned that artificial intelligence tools are “prone to errors” and should not be blindly trusted, in an interview with the BBC. He emphasized that AI models, including Google’s Gemini 3.0, are best used alongside traditional information sources such as Google Search.
Pichai said generative AI is useful for creative tasks but stressed that users must understand the limitations of these systems, particularly for sensitive topics such as health, science, or news. He acknowledged criticism of AI-generated summaries and said Google is investing in AI security and transparency, including open-sourcing tools to detect AI-generated images.
The comments come as Google integrates its Gemini chatbot into AI Mode for Search, marking what Pichai called a “new phase of the AI platform shift.” He highlighted the balance between rapid AI development and implementing safeguards, noting that no single company should control AI technology given its potential impact.
Experts have warned that generative AI can produce misleading or false outputs, and Pichai underscored that Alphabet aims to be “bold and responsible at the same time” while rolling out its AI products.
