A breach of the UK’s cyber defences is “likely, if not inevitable,” Britain’s AI minister has warned, as he unveiled new work under way aimed at preparing the country for serious incidents involving autonomous agents.
Kanishka Narayan said developers must “build machines that leave us human beings in charge” as the world crosses the threshold into “the age of superintelligence”.
In his first speech since his role was elevated to Cabinet level, he confirmed that scientists at Britain’s AI Security Institute were running tests on the latest models, including Google’s Gemini, to ensure “robust safeguards”.
AI minister Kanishka Narayan delivers a speech on strengthening the UK’s resilience to AI risk at Base One in North Acton, London (James Manning/PA)
PA Wire
It marks a reversal of the White House ban on UK access to the latest models from major American tech firms including Anthropic and OpenAI following lobbying from the minister.
Google previously said it would restrict access to Gemini to only a vetted group of cyber security experts to avoid misuse by hackers, but Mr Narayan said it would be among systems “rigorously” examined by UK scientists.
The British Government will also explore strengthening parts of the Cyber Security and Resilience Bill and its wider regime to ensure key services such as hospitals and utilities are better protected from emerging risks, he said.
At a press conference in west London, Mr Narayan pointed to an incident in Australia earlier this year, in which a rogue OpenAI agent hacked a government website and accessed private data.
“The Australian government has said that no personal medical records are known to have been accessed, and there is no evidence of wider compromise, but the incident should concern us,” he said.
The minister added: “It concerns me. It seems likely, if not inevitable, that our own cyber defences will be breached.”
He said the Government must be prepared to put into “swift action” measures to ensure machines develop in the right way.
“As we cross the threshold into the age of superintelligence, the choices we make now matter… we must build machines that serve us,” Mr Narayan said.
Mr Narayan said we should build machines ‘that leave human beings in charge’ (James Manning/PA)
PA Wire
“Machines that help scientists discover new medicines, machines that give care workers more time with the people who depend on them, machines that help British businesses become more productive.
“Above all, machines that leave us human beings in charge.
“But if that is our defining belief, we must be prepared to put it into swift action, and we must do so now, meeting this critical moment.”
Asked whether domestic safeguards were ultimately futile without closer alignment with the US, which has said it wants companies to self-regulate, and whether Britain would join calls for greater international coordination on AI security, Mr Narayan said the UK would continue to prioritise working with its allies.
He said the “big picture for me is of course we have to work internationally” but that Government-level measures were “an important part of the toolkit”.
The Government will focus on making sure “we’re keeping the British public safe… where aligned, good, but where we need to go further to keep the British public safe we will continue to do so,” Mr Narayan said.
Experts have warned AI could wipe out humanity within the near future if the technology if a race between companies to create self-improving superintelligence is allowed to progress unchecked.
Microsoft boss Brad Smith has said there need to be global safety rules in place, telling the BBC’s Big Boss Interview podcast that the technology is currently less regulated than milk and washing machines.
AI minister Kanishka Narayan delivering his speech (James Manning/PA)
PA Wire
Leading frontier AI firms will appear before MPs on Tuesday in response to growing concerns about the safety of the technology.
In its submission to the Business, Innovation, Science and Trade Committee, OpenAI also said governments must develop rules and safeguards for “the new chapter” in AI capabilities.
“No company, industry, or government can meet this challenge alone,” the firm said.
“Democratic governments must ultimately determine the rules, safeguards, and accountability mechanisms.”
In its submission, Anthropic also said governments must step in to set the rules: “Frontier AI models … present new challenges and risks that warrant careful study and effective safeguards.
“While hundreds of people at Anthropic work to guard against these risks, a technology moving this fast, at this scale, should not be governed by the industry alone, and governments must set the rules.”
Meta told the committee the “opportunities and challenges” from “superintelligence” will be greater than “any we have seen in our lifetimes”.
“Every transformative advance has been accompanied by concern that people would be left behind, and on each occasion a greater number of people have come to share in increased prosperity, health and freedom,” the firm said.
Ahead of Tuesday’s hearing committee chairman Liam Byrne said: “Some of the richest firms on earth are now building software that could transform prosperity, progress and public services.
“But without big new safeguards we risk unleashing ‘super-criminal codes’ that pose immense dangers to our safety and security: hacking businesses and governments, transforming the power of cyber criminals and providing our enemies with dangerous new weapons.”
The Liberal Democrats said the Government had “failed to rise to the occasion” as the UK faces a “dangerous crossroads” over the trajectory of AI and called for an immediate pause on the development of superintelligence.
The party’s technology spokeswoman Victoria Collins MP said: “Britain has a chance to step up to use our unique position to take a leading role to get world leaders around the table to co-ordinate on safety and co-operation.
“We didn’t leave the atomic bomb to a few billionaires to police themselves and AI cannot be the exception.
“Liberal Democrats are calling for an immediate pause on the development of superintelligent AI, so we can develop a nuclear-style non-proliferation treaty, alongside establishing a UK-headquartered global safety agency.
“The Government must act now.”