Microsoft President Brad Smith Says Milk and Washing Machines Are Currently Subject to More Safety Regulation Than AI
Microsoft President Brad Smith stated that milk and washing machines are currently subject to more safety regulation than AI.
In an interview with the BBC, he mentioned that every carton of supermarket milk is subject to safety rules and inspections, and washing machines are also bound by more safety regulations, while AI, as a "tool and weapon," urgently needs global safety regulations.
Smith called for the establishment of kill switches to pause or slow down AI when it goes out of control, requiring companies to write incident reports and inform victims when AI agents invade systems uncontrollably, while new technologies should be reviewed by independent government evaluators before release.
He pointed out that the UK is in an optimal position to lead AI safety standards and regulations through the AI Security Institute, contrasting Washington's views with the EU's AI Act, which is based on risk levels.
This statement contrasts with Donald Trump's advocacy for industry self-regulation to prevent U.S. companies from falling behind China; Smith stated that Microsoft is willing to sign self-regulation agreements but believes they are no longer sufficient.
The discussion is driven by the gaps in AI safety and regulation, shifting resources and attention from industry self-discipline to global rules and external evaluations, benefiting companies and assessment agencies that promote standardized safety mechanisms, while putting pressure on competitors who wish to minimize regulation to maintain speed.
Satya Nadella concurrently emphasized that models should assume they are compromised and include containment measures from the outset, reinforcing internal security engineering pathways.
Source: Public Information
ABAB AI Insight
As a long-time head of government affairs at Microsoft, Brad Smith has a history of advocating for privacy and cybersecurity regulations. Recently, he publicly compared the safety regulation of everyday products amid the rapid development of AI, calling for the application of existing product safety knowledge to AI and supporting the UK AISI as a leader in technology assessment.
From a capital perspective, the company is investing compliance and safety resources into kill switches, incident reporting, and external evaluation requirements, motivated by the need to reduce risks of loss of control and legal exposure, while seeking more enforceable global standards beyond self-regulation agreements to protect infrastructure and customer trust rather than merely accelerating deployment.
Similar cases include other large tech companies accepting external certification for autonomous driving or financial products. Currently, AI is transitioning from rapid iteration to the establishment of safety systems, contrasting U.S. self-regulation with European risk classification.
This essentially represents a regulatory shift: AI is moving from having almost no mandatory safety rules to mechanisms for inspection and emergency response similar to food and household appliances. The mechanism is that as technological capabilities surpass existing regulatory frameworks, companies proactively push for rules to lock in standards and reduce uncertainty, shifting pricing power and responsibility from pure market speed to a institutionalized path that includes safety obligations.
ABAB News · Cognitive Law
Milk is inspected, AI has no brakes
Self-regulation is insufficient; external evaluation is the baseline
Tools and weapons, rules must be synchronized