The UK plans to make artificial intelligence a central issue during its 2027 Group of 20 (G20) presidency, as the White House opposes global governance of the technology.
In September, Prime Minister Andy Burnham told the United Nations General Assembly that AI would be a central priority for the UK presidency and that Britain would pursue shared principles and standards. In February, Michael Kratsios, director of the White House Office of Science and Technology Policy, said, “We totally reject global governance of AI.”
A joint statement hosted by the Netherlands government urges companies to test AI systems before deployment and arrange independent evaluations. It calls on governments and regional bodies to coordinate standards and improve transparency, including through shared reports of serious safety incidents. The statement also asks U.N. member states to explore an international institution. It remains open for endorsements and is not an adopted treaty.
The Guardian’s Oct. 22 editorial cited a recent PHP security incident in arguing that international AI safety efforts should continue despite U.S. opposition. PHP maintainers released fixes Oct. 21 for CVE-2026-109618 and CVE-2026-109619, flaws that can let attackers run code remotely without authentication. PHP said systems handling web requests are at risk, particularly those using Apache’s mod_php. The fixed releases are PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35; PHP advised users to install a fixed version promptly.
OpenAI credited Calif, working with GPT-5.6 Sol, with identifying a critical flaw and said Wiz had independently found both vulnerabilities. Wiz said Gemini 4 Argon contributed to its research. PHP’s advisory did not identify the research teams or tools.
Amazon Threat Intelligence said its MadPot honeypot recorded an exploit attempt on Oct. 17, four days before the public disclosure. In its account of the activity, including later attempts, Amazon said it could not determine whether any would have worked against real systems or how the actor learned of the flaw in advance. Amazon also warned that newer analysis tools could shorten the time attackers need to develop exploits once technical details are available.
The UK government said its AI pilots identified and addressed 1,352 vulnerabilities across 548 code repositories in 18 departments. The 2026 International AI Safety Report said decision-makers have few quantitative measures for AI risks and face substantial evidence gaps. Burnham has pointed to the G20’s response to the 2008 financial crisis as a precedent for countries agreeing on principles they can apply at home.
The Cyber Security and Resilience Bill is due for a report-stage debate in the House of Lords on Oct. 26. The UK says it will announce its G20 presidency priorities in early 2027, ahead of the leaders’ summit scheduled for Nov. 27-28, 2027.
