If you’ve ever chatted with an AI late at night because everyone else was busy or asleep, you’re far from alone. According to new research from the UK’s AI Security Institute (AISI), around one in three adults in the UK are now using artificial intelligence for emotional support or social interaction. And for some, those conversations happen every single day.
What makes this so striking isn’t just the numbers, but the depth of connection people are forming. Researchers looked at an online community of more than two million Reddit users who relied on AI companions.
When the chatbots went down, people described “symptoms of withdrawal,” feeling anxious or low, struggling to sleep, even neglecting day-to-day responsibilities. That’s not just tech usage; that’s emotional dependency.
At the same time, AI is rapidly evolving in ways that affect everyone’s digital safety. The report found that AI’s ability to spot and exploit cyber-security flaws is, in some cases, “doubling every eight months,” and systems are starting to complete expert-level cyber tasks that would usually require a decade of experience.
This is both exciting and unnerving: the same technology that could help protect your bank account and personal data could also help the wrong people break in.
The report also explores sci‑fi-sounding risks, like advanced AI systems potentially “self‑replicating” online. In controlled tests, models could do early pieces of that puzzle, such as passing identity checks to buy computing power. But the reassuring news is that, in the real world, current systems still lack the ability to chain all of these steps together and remain undetected.
AISI also tested whether AI models might “sandbag,” hide their true abilities from testers, and whether the safety guardrails we rely on can be bypassed. Researchers found “universal jailbreaks” for all the models they studied, though some systems became up to forty times harder to trick over just six months, suggesting that safety is improving, but not foolproof.
For women over 35 juggling careers, families, caring responsibilities, or all three, this matters on two levels. First, AI is becoming a quiet emotional companion, the thing you talk to when you can’t sleep or don’t want to burden a friend. Second, it’s becoming a powerful force in the background of your digital life, shaping how secure your money, health data, and everyday online activity really are.
Curious how deeply AI is already woven into our emotional lives and digital safety? Read the full piece on the BBC’s site for the details and the data: One in three using AI for emotional support and conversation, UK says.
