What happened
In a BBC Big Boss Interview, Microsoft president Brad Smith argued that AI is advancing faster than safety safeguards and that the UK should lead in establishing global AI rules. He proposed three specific measures: an emergency brake for critical AI systems, mandatory incident reporting, and government pre-release certification for frontier models.
Brad Smith, president of Microsoft, appeared on the BBC's Big Boss Interview to discuss the current state of . He stated that artificial intelligence is advancing faster than the safeguards surrounding it and that industry self-regulation is 'not sufficient any longer.' Smith emphasized that the world should be more concerned about AI safety than any previous digital technology in his 33 years in the industry.
Smith outlined three specific measures for improving . First, he proposed a 'safety brake or emergency brake' for AI systems controlling autonomous devices or critical infrastructure, suggesting this kill switch should be integrated into both the companies running the models and the cloud infrastructure. Second, he called for a legal obligation to report incidents where AI agents 'break loose' and compromise systems, similar to data breach reporting requirements.
The third and most significant measure Smith proposed was government pre-release certification for frontier models. He compared the current lack of regulation to the fact that washing machines, milk cartons, and passenger aircraft are subject to strict safety rules and inspections. Smith argued that government evaluators should have to sign off on frontier models before they are released to the public.
Smith specifically identified the United Kingdom as the country best placed to lead these efforts. He praised the UK AI Security Institute for having the strongest technical capability in the world to test AI models. He noted that the US began building similar capabilities but slowed down, and he described Britain as a 'bridge across the Atlantic and across the English Channel,' suggesting it could mediate between US and European regulatory approaches.
In the interview, Smith also addressed Vice President JD Vance's accusations that Microsoft abuses the H-1B visa system. Smith refuted these claims, stating that the VP's staff 'got the maths wrong' and that 80% of Microsoft's 6,000 visa applications were for existing employees. He also stated that the company hired 9,000 Americans last year and that the cost of new grid capacity for data centers should be borne by the industry, not neighbors.
Why it matters
This statement represents a significant shift in corporate stance, with a major AI developer explicitly calling for government intervention and pre-release regulation rather than relying on self-regulation. By identifying the UK as the ideal leader, Smith highlights the potential for the UK AI Security Institute to set global standards, which could influence how frontier models are deployed worldwide and impact the regulatory landscape for companies like Microsoft and OpenAI.
This interview marks a notable departure from the 'tremendous self-regulation' praised in recent White House AI accords. By explicitly calling for government intervention and pre-release certification, Smith is advocating for a regulatory framework that could fundamentally change how frontier AI models are developed and deployed.
The proposal for the UK to lead global rules is significant because it positions the UK AI Security Institute as a potential global standard-setter. This could influence international regulatory trends and provide a model for other countries to follow, potentially creating a more unified global approach to AI safety.
The call for mandatory incident reporting and emergency brakes addresses growing concerns about the autonomy and reliability of AI systems. These measures could provide greater accountability and transparency for AI developers and users, helping to build public trust in AI technology.
Smith's comments also highlight the tension between rapid AI development and the need for robust safety safeguards. As AI systems become more capable and integrated into critical infrastructure, the potential risks of failure or misuse increase, making the need for effective regulation more urgent.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').Why can ethical evaluation not be reduced to one model score?
What to watch next
Watch for the UK government's response to these proposals, particularly regarding the AI Security Institute's role in pre-release testing. Also monitor if other major AI developers adopt similar calls for mandatory government sign-off or if this leads to new legislative proposals in the UK or US.
The UK government's response to Smith's proposals will be a key indicator of whether the UK is willing to take on a leadership role in global . Watch for any new legislative proposals or regulatory actions from the UK government in the coming months.
The reaction from other major AI developers, including OpenAI, Google, and Anthropic, will be important to monitor. Will they support Smith's calls for government pre-release certification, or will they continue to advocate for self-regulation?
The implementation of the UK AI Security Institute's testing capabilities will be crucial. If the institute is able to effectively test and certify frontier models, it could become a model for other countries to follow. Watch for any new testing results or certifications from the institute.
The ongoing debate over and regulation in the US will also be important. Smith's comments may influence the US regulatory landscape, particularly as the US government continues to develop its own AI safety capabilities.