The Artificial Intelligence operating without human control could one day pose a security threat, the head of Britain’s MI5 spy agency said today, although he noted that he does not foresee any “Disclosure” scenario like those depicted in science fiction movies.

Technology leaders have warned of the dangers of runaway Artificial Intelligence, saying that people should be concerned about a future similar to that of the Extremist film series, in which an AI system gains self-awareness and attempts to eliminate humanity.

In his annual speech on threats to Britain, MI5 chief Ken McAllum said AI was being used by British security services to make their work more effective, while terrorists were using it for propaganda and target identification, and state agents were using it to manipulate elections and increase government attacks.

“But in 2025, while we address today’s threats, we must also overcome the next frontier: potential future risks from non-human, autonomous AI systems that may escape human oversight and control,” he said.

Macalum noted that he “chooses his words carefully” because there is a risk that what he says could be misinterpreted and seen as overly destructive. I don’t foresee Hollywood-style scenarios (…) but, as the capabilities of the TF continue to grow, one would expect organisations like MI5 (…) to think deeply, starting today, about what the Defence of the Realm should look like in the coming years,” he said.

“AI may never mean anything bad for us. But it would be imprudent to ignore its potential to cause harm. We’re watching it,” he added.

> More World