Generative AI Exposes Private Numbers, Fueling Scams for Millions of Britons
Updated
Updated · The Independent · May 10
Generative AI Exposes Private Numbers, Fueling Scams for Millions of Britons
4 articles · Updated · The Independent · May 10
Google’s AI and other chatbots are surfacing private phone numbers in answers, triggering repeated harassment for some victims and directing other users to scammers posing as businesses.
LLMs generate those replies from scraped web data, outdated records and even placeholder details; researchers say criminals also poison reviews, comments and other pages so fake support numbers get absorbed into AI systems.
Virgin Media O2 said millions of Britons have already been shown fake customer-service numbers via AI tools, creating openings for data theft and fraud when callers think they are reaching trusted brands.
Google said it has safeguards and removal tools for personal content, but one victim said a legal privacy request to blacklist their number drew no response and the calls continued daily.
Privacy and security experts say the problem could worsen as newer models train on more data, while victims have few remedies because AI systems cannot easily "forget" information once learned.
With criminals poisoning AI, can we ever trust the answers our smart assistants give us?
If AI can't forget, is our private data now permanently exposed online?
Who is legally responsible when an AI lies or leaks your personal secrets?
2026 UK AI Fraud Crisis: Scale, Impact, and the Fight Against Deepfake Scams
Overview
In 2026, the UK faces a rapidly escalating AI fraud crisis, with sophisticated scams and deepfakes causing significant financial losses and impacting government, businesses, and citizens. The widespread use of AI-driven threats has shifted from a speculative issue to a real and urgent danger, challenging traditional security measures and making it hard for both technology and people to tell what is real. As these fraudulent activities grow in scale and complexity, existing detection tools and human judgment often fail to keep up, highlighting the urgent need for stronger defenses and greater awareness.