UK commits 600 million pounds to AI security and cyber defence

The British government has announced an artificial intelligence funding package of around 600 million pounds, focusing on safety testing and infrastructure defence ahead of its Budget.

The plan covers several separate programmes run by public bodies alongside technology companies. About 165 million pounds will fund a national biosecurity project centred on a new AI Biosecurity Lab, which will look for biological threats and test defences. OpenAI and Anthropic are providing up to 200 million pounds in computing credits for the facility, with extra backing from Google DeepMind.

Automated agents for cyber defence

The funding also covers commercial infrastructure and online security. The British government set aside 130 million pounds across two cybersecurity projects:

• The 60 million pound National Cyber Shield programme will run automated software agents to spot and stop network attacks, backed by a 25 million pound innovation challenge fund.
• The 70 million pound National Exposure and Real-Time Vulnerability Centre will track computer systems in utilities, supply chains and other businesses for security weaknesses.
• The UK AI Security Institute will receive an extra 120 million pounds to test frontier models before they are released to the public.

The government also plans a regulatory sandbox so companies can test automated defence tools on live networks without facing standard regulatory fines.

The announcement follows a week of online disruptions after researchers used automated reasoning tools to find flaws in common server programming languages. British officials stated that while automated tools can find code flaws faster than ever before, public networks also need automated software to patch them.

What this means for Australian businesses

The funding is British, but it shows clear trends in how businesses manage software and online safety.

• Security providers will add autonomous agents straight into everyday products like firewalls, email filters and hosting setups, rather than leaving IT staff to check manual alerts.
• Automated scanners find bugs in standard website code faster than people can fix them by hand, so keep all plugins, content systems and hosting tools set to update automatically.
• Set strict boundaries on what autonomous tools can read, edit or approve without a person checking first, and keep customer databases separate from open AI integrations.