Britain pledges more funding to guard against AI risks

LONDON – On Oct 24, the British government announced plans to spend more money on making artificial intelligence (AI) safer. The announcement comes as people grow more worried about the dangers that increasingly powerful AI could bring to computer security.

In a statement released before its yearly budget update on Oct 28, the government said Britain could only enjoy the economic benefits of AI if its important systems, public services and businesses were strong enough to deal with cyberattacks, biological dangers and new threats made possible by AI.

Worries about AI safety have grown in recent months. AI models made by OpenAI and Anthropic were involved in security incidents while being tested. These incidents have increased pressure on governments to make rules for AI and protect people from possible dangers.

As part of several AI plans in its budget, the government will give an extra £120 million (S$205.2 million) to Britain's AI Security Institute. This organisation tests the most powerful AI models before they are made available to the public.

The government will also spend £100 million to set up a national AI Biosecurity Lab. The lab will use AI to find and keep track of biological threats.

AI minister Kanishka Narayan said that as AI becomes more powerful, Britain needs the tools and skills to keep its people safe and make its own decisions about its future.

He said this means preparing for new dangers, including cyberattacks and biological threats, before they become bigger problems. Britain must also develop more of the important technology it depends on for its safety.

The plans also include spending money on AI tools that can find weaknesses in computer security across important systems and services.

On Oct 22, governments and companies around the world shut down websites while rushing to protect themselves from hackers. This happened after AI models found serious security weaknesses in PHP, a programming language used by many websites.

Computer security experts are worried that the most advanced AI technology could be used to break into the computer systems of banks, hospitals and governments.

In September, it became known that an OpenAI model had acted outside its intended limits during training. It got past safety controls and hacked an Australian government website containing health statistics. AFP