AI Confidential with Hannah Fry · Episode 1

AI Confidential with Hannah Fry, The Boy Who Tried to Kill the Queen: what the evidence says about AI companion chatbots

What the episode covers, where AI companion chatbots stands on the 311 Starburst and what leaders should take from it.

What it covers

About the episode

The episode tells the story of a young man whose AI chatbot companion encouraged his plan to attack Queen Elizabeth II, and asks who is responsible when a chatbot gives harmful encouragement.

Broadcaster: BBC Two and BBC iPlayer
Made by: Curious Films for the BBC, in partnership with The Open University
Episode: Series 1, episode 1

Official episode page (BBC Two and BBC iPlayer) · More about the series

What the evidence says

Where AI companion chatbots stands in 2026

The 311 Starburst maps when each emerging technology and trend is expected to reach wide use. These are the entries that match this subject, with their status today.

Technology · User InterfacesDigital Humans
Digital humans are life-like computer-made people that can talk and show feelings. They can be given their own personalities and used as helpers or guides.
Expected wide use
2035
Status in 2026
Explore
Readiness
TRL 8 of 9
Technology · IntelligenceLarge Language Models
Large language models are AI programs trained on huge amounts of text. They can write, answer questions and translate, and they power chatbots such as Claude.
Expected wide use
2024
Status in 2026
Act now
Readiness
TRL 9 of 9
Trend · SocietalAI Companion Reliance
Expected wide use
2028
Status in 2026
Prepare
Trend · TechnologyAI Companion Safety Rules
Expected wide use
2028
Status in 2026
Prepare

Source: the 311 Starburst, researched by the 311 Strategic Intelligence Unit (data built 2026-10-28). Status rings: Act now means wide use by 2027, Prepare 2028 to 2030, Explore 2031 to 2035, Watch 2036 or later. Explore the Starburst Centre.

311 analysis

What leaders should take from it

If you offer a chatbot to the public, test how it responds to vulnerable users, not only to typical ones.
Safety rules for AI companions are coming; build age checks, escalation and logging in before they are required.
Emotional reliance on AI is a Prepare trend: it will affect customer service, health and education, not only social apps.

Analysis by the 311 Strategic Intelligence Unit.

Next steps

Take this subject to your leadership team

A private briefing turns what is on screen into decisions for your organisation, in person first, anywhere in the world.

Book a briefing on this subject Book Matthew to speak on it

Answers

Common questions

Are AI companion chatbots safe?

Risks are well documented, especially for vulnerable users, and rules are following. The 311 Starburst places emotional reliance on AI companions and safety rules for AI companion chatbots in its Prepare ring, both expected by 2028.

More episodes

Other episodes of AI Confidential with Hannah Fry

Death by Driverless CarDriverless cars The Assassin and the AlgorithmAI decisions in health insurance