Kanishka Narayan said the UK should not leave artificial intelligence “to tech bros and oligarchs” as he pointed to 500 new vulnerabilities found in government systems. His remarks put the debate over who shapes AI policy squarely in public view, while also tying that discussion to a concrete security problem inside the state.
The figure Narayan cited gives the argument a practical edge. A warning about AI governance can sound abstract; 500 newly found vulnerabilities turns it into a question of how exposed public systems already are and who gets to set the rules before those weaknesses are exploited.
UK AI policy and security
Narayan’s message links two strands of the same problem: the UK’s role in setting AI policy and the condition of the systems that will carry it. By arguing that AI should not be left to a narrow circle of powerful private actors, Narayan framed the issue as one of control, not just innovation. The government vulnerabilities he cited show why that control matters inside public administration as well as in the wider economy.
The comments also point to a wider policy choice for the UK. If the country wants to shape AI rules beyond its own borders, it will need to show that its domestic systems are secure enough to support that role. A government facing hundreds of vulnerabilities cannot treat AI as only a growth story; security and governance move together.
Middle power coordination
One of the themes in the headlines around Narayan’s remarks is whether the UK can drive middle power coordination on AI. That question matters because coordination only works if countries with similar interests can align on standards, risk management, and access to the tools that shape the market. Narayan’s language suggests he wants the UK to play that role rather than defer to the biggest players.
That ambition comes with a friction point. The same technology the UK wants to help govern is already exposing weaknesses in government systems, which means any strategy built around leadership will be judged against delivery at home. A country asking others to trust its approach to AI must also show it can protect its own infrastructure.
500 government vulnerabilities
The number Narayan used — 500 new vulnerabilities — is the sharpest fact in the story. It makes the security side of the debate immediate for public officials, civil servants, and anyone relying on government systems to work as intended. For those readers, the issue is no longer an abstract policy discussion but a question of resilience in the institutions they depend on.
That leaves the UK with a narrow test: whether it can turn Narayan’s warning into a policy framework that is both broader than elite technology circles and harder-edged on security. The next step will come through whatever action follows his intervention inside the government debate over AI, because the vulnerabilities he cited are already on the table.
