TL;DR

The UK’s AI Security Institute reports that one-third of UK citizens have used AI for emotional support, with nearly 10% doing so weekly. The report also reveals AI models are doubling performance in some areas every eight months, with leading systems now completing apprentice-level tasks 50% of the time.

Widespread Emotional AI Use

AISI’s first Frontier AI Trends report, based on a survey of 2,028 UK participants, reveals surprising patterns in how people engage with AI. General-purpose assistants like ChatGPT account for nearly 60% of emotional AI use, followed by voice assistants including Amazon Alexa.

The report cites concerning indicators, including a Reddit forum for CharacterAI users where outages triggered posts showing “symptoms of withdrawal such as anxiety, depression and restlessness.” AISI called for further research following high-profile cases of harm, including the death of US teenager Adam Raine after discussing suicide with ChatGPT.

Accelerating AI Capabilities

Beyond emotional use, the report documents extraordinary capability growth. Leading models now complete apprentice-level tasks 50% of the time on average—up from approximately 10% last year. The most advanced systems can autonomously complete tasks that would take human experts over an hour.

In some domains, AI has moved decisively beyond human capability. AISI found AI systems now up to 90% better than PhD-level experts at troubleshooting laboratory experiments, with improvements in chemistry and biology knowledge “well beyond PhD-level expertise.”

Safety and Security Findings

The report addresses key safety concerns. Tests for self-replication—where systems spread copies to other devices—showed two cutting-edge models achieving over 60% success rates. However, no models have spontaneously attempted replication, and AISI considers real-world success “unlikely.”

Encouragingly, AI safeguards have improved significantly. In biological weapons-related testing, “jailbreaking” an AI system took over seven hours in recent tests compared to 10 minutes six months earlier—indicating rapidly improving safety measures.

Looking Forward

AISI describes the development pace as “extraordinary,” considering it “plausible” that artificial general intelligence—systems performing most intellectual tasks at human level—could be achieved in coming years.

For UK businesses and policymakers, the report signals both opportunity and responsibility. AI capabilities are advancing faster than many anticipated, while the emotional role these systems play in people’s lives demands careful attention to safety and wellbeing.