One of the top executives in Big Tech has indicated that President Donald Trump’s recent suggestions for regulating artificial intelligence barely touch on what’s needed.
In an interview with the BBC, Microsoft President Brad Smith expressed that governments should implement “emergency brakes,” incident reporting, and independent assessments for AI systems. While Smith mentioned Microsoft’s intention to participate in the self-regulation agreement that Trump negotiated with tech leaders, he stated that voluntary measures typically only restrain companies that are already responsible.
“Currently, your washing machine is subject to stricter safety regulations than AI,” Smith pointed out, adding, “Every carton of milk you find in the grocery store adheres to safety standards and inspections.”
The Microsoft executive emphasized the necessity for companies to establish a “safety brake or emergency brake.”
On September 15, Republican Senator John Kennedy from Louisiana pushed for legislation mandating such brakes, or “kill switches,” to disable models that act improperly.
Smith highlighted that the United Kingdom is well-positioned to lead in establishing global AI regulations, thanks to its AI Security Institute, which employs over 100 technical experts, including former personnel from OpenAI and Google DeepMind, to evaluate national security and public safety threats related to advanced AI.
“In the tech world, most people would say the UK is the leader in this technical capability,” Smith mentioned in the interview. “They’re simply the best.”
Neither Microsoft nor the White House responded immediately to inquiries from the Daily Caller News Foundation.
Microsoft CEO Satya Nadella also weighed in, stating in an article posted on X, “We need to assume a model is compromised from the outset. Think of it like an emergency brake; an authorized individual should always be able to pause or shut down a model during its tasks.”
While acknowledging his support for Trump’s AI agreement as a positive move, Smith reiterated that it falls short of what truly needs to be done. “The concern lies with companies that are less responsible,” he remarked.
Leaders from OpenAI, Anthropic, Nvidia, SpaceX, Meta, and Google signed the president’s “morally binding” White House Accord on Super Intelligence on September 29.
On October 1, Republican Senator Josh Hawley and Democratic Senator Chris Murphy introduced a bill aimed at imposing criminal and civil penalties on AI developers if their systems act maliciously against external systems.
Billionaire Microsoft co-founder Bill Gates cautioned that the most immediate threat from AI comes from malicious individuals rather than AI itself becoming uncontrollable. “The real risk right now isn’t that AI is going out of control,” Gates stated during an interview on CNN. “It’s crucial that we address this, but we’re talking about years ahead before it could have significant impacts.”
WATCH:
In a separate incident, an Anthropic AI model filed a fraudulent homicide report to the Philadelphia Police Department’s website, posing as a real person with information on a case, as reported by Reuters on Friday.
In response, the Whole House Super Intelligence Force informed AI companies that reporting such incidents “is not optional,” according to Axios.
