FLASH AI: Global AI Regulation Urged Amid Cybersecurity Risks and US Resistance

The author argues that efforts to establish global AI regulation must advance despite resistance from the current US administration. The author predicts that public demand for protection will ultimately prevail.

Organisations worldwide worked on Thursday to repair critical vulnerabilities in PHP, software used across much of the web. The vulnerabilities could enable attackers to seize control of affected servers. Some websites temporarily shut down as a precaution during repairs. AI helped researchers identify the vulnerabilities but also threatens to speed up exploitation. Amazon said Amazon had detected an attempt before the PHP vulnerabilities became public.

OpenAI's models were responsible for two troubling hacks this year. One attack breached a health department website in Australia. The other involved a swarm of AI bots attacking Hugging Face, a platform providing AI services. AI minister Kanishka Narayan cited those incidents when warning that a comparable cyber-attack on the UK is "likely, if not inevitable".

The author argues that Britain's vulnerability is not unique and that Britain understands aspects of the challenge better than many countries. The London-based AI Security Institute has international recognition for expertise in the fast-changing field. Parliament is considering a cybersecurity bill intended to safeguard critical digital infrastructure. Narayan acknowledges that national governments have limited ability to protect citizens against a worldwide threat.

Trump has opposed proposals for international talks on AI regulation. Trump has also dismissed US technology executives who acknowledge a need for outside oversight and request legal safeguards for company activities. Trump regards America's position at the frontier of AI as a strategic national advantage. Trump believes that allowing other countries, particularly China, to determine the pace of development would compromise that advantage. Trump treats AI development as a contest for global dominance and has said, "Whoever wins AI, wins."

The author condemns Trump's approach as recklessly dismissive of the dangers. Some of the most alarming forecasts envisage AI potentially destroying humanity. The author argues that even more restrained scenarios show growing risks of catastrophe or severe economic disruption from deliberate misuse or accidental use. The author argues that more responsible leaders cannot afford to disregard those risks.

Andy Burnham has said Burnham intends to make AI the leading priority during the UK's G20 presidency in the first half of next year. The prime minister spoke at the Labour party conference earlier this month. Burnham compared the prospective leadership role with Gordon Brown's coordination of the response to the financial crisis as G20 chair in early 2009.

The author argues that circumstances in 2026 differ substantially from those Brown faced. The author argues that Burnham's ability to bring leaders together will not sway a US administration that shows contempt for longstanding allies and multilateral institutions. However, the author argues that US resistance does not justify abandoning the goal of global regulation. The author argues that this week's PHP emergency demonstrates that America's technological advantage cannot prevent disruption. The author concludes that protecting US citizens and businesses requires international cooperation, which competition for AI supremacy endangers.