TL;DR

A fifth of UK adults (20.6%) now consult AI about personal matters or relationships, according to Opinium polling for the Social Market Foundation (SMF), released on 22 October. Among Gen Z the figure is 41.7%. More than half of those who use AI this way (53.7%) rated it above the friends and relatives they might otherwise ask. The think tank wants a package of safeguards, starting with a right to be told when you are talking to an AI.

What the poll found

Opinium surveyed 2,000 UK adults between 16 and 20 October 2026, weighting results to be nationally representative. Beyond the 53.7% who preferred AI, three in ten of the people using it for personal advice rated it as helpful as the people around them.

The poll comes with The Imitation Game, an SMF briefing on AI that can convincingly mimic empathy and intimacy. SMF director Theo Bertram said people “can no longer tell the difference in short text-based conversation”, and are using AI to rehearse difficult conversations and look after their wellbeing.

The risks SMF sets out

The think tank points to fraud, which on its figures is now 45% of crime in England and Wales, and which more convincing AI impersonation could worsen. It cites laboratory studies finding AI “reliably more persuasive than expert humans”. It also coins an “authenticity tax”: the time and money spent showing that a person, or a piece of content, is genuine. For teenagers, it warns of “cognitive offloading”, where relying on AI to read relationships means missing the chance to work them out themselves.

What SMF wants

Its seven recommendations are:

  • disclosure: anyone talking to an AI rather than a person should be informed, mirroring an existing EU rule
  • a guaranteed, affordable human option in high-stakes services such as welfare, justice and health
  • baseline rules on watermarking AI output and tracing where it came from
  • deepfake provisions in election law, following Canada, backed by enforcement powers
  • safety duties built into AI design, such as spotting when a user is in crisis
  • a prohibition on using emotional pressure to keep people engaged or push sales
  • protections suited to children’s ages, and further study of AI’s effect on young people’s relationships

Looking forward

SMF says parliamentarians will soon face decisions on chatbots, mental health and AI in political campaigning, an area where Bloomberg reports some UK politicians already use AI focus groups. In our view, the “right to know” is the recommendation most likely to gain traction, because it already has an EU model. The poll’s headline needs care, though: the 53.7% figure covers only the fifth of adults who use AI for advice, not the public as a whole.

Share this article