What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
U.S. Treasury Secretary Scott Bessent says people at AI labs must take responsibility for the risks their systems create—and that warnings about those risks need to be matched with practical solutions. His position is to keep developing AI while preparing for harm: what he called “safe acceleration.”
What did Bessent say about AI risks?
In an interview reported by Axios, Bessent said responsibility rests with the people building AI systems. “It is the people at the labs who have to accept responsibility, and I agree with that,” he said. He also criticized warnings that offer no path to address the dangers, arguing that alarm without solutions is not leadership. The Economic Times reproduced his remark that “alarmism without solutions by some of the AI community” is not leadership.
That criticism does not amount to a claim that AI risks are imaginary. Bessent said the United States should prepare for serious incidents and give more attention to resilience and defense, alongside the development of advanced models.
Who does Bessent say is responsible when AI systems cause harm?
He pointed to people at AI labs—the organizations developing the systems—as responsible for addressing risks. The remark is a call for accountability, not a description of a specific legal standard, enforcement mechanism, or liability ruling. The interview report does not establish which individuals or organizations would be legally liable in a particular case.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minute#1 Best Overall
What does “safe acceleration” mean in this context?
Bessent used “safe acceleration” to describe pursuing AI development while preparing for its potential dangers. In the same interview, he said executives who believe they could lose control of increasingly powerful models should slow down: “Well, then they should slow down.” Together, those comments set a conditional boundary: continued progress is not a reason to ignore safety concerns, and a credible loss-of-control concern should prompt caution.
He also urged more focus on resilience and defense. In practice, that means treating preparation for incidents as a complement to frontier-model development, rather than assuming development alone will make systems safe. The interview report does not specify a technical checklist or define a measurable threshold for when a lab should pause.
Rank #2
What AI incident-notification proposal did the U.S. discuss with China?
Axios reported that Bessent discussed a proposed U.S.-China process for communicating about serious AI incidents. The reported concerns included uncontrolled AI agents and non-state actors using AI for cyber or biological threats. The proposal was described as a way to communicate about incidents—not as an agreement already in force or an operational notification system.
The same reporting said U.S. model reviews were voluntary at the time. It also said the administration reserved the right to intervene if a lab proceeded despite safety concerns. Those points describe the reported policy context; they do not establish that a particular intervention occurred or that the bilateral channel was later implemented.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →How does this fit with Treasury’s work on AI in finance?
Treasury has addressed AI risks in financial services separately from Bessent’s reported proposal for U.S.-China incident communications. The distinction matters: a domestic financial-sector initiative is not evidence that a bilateral incident-notification channel exists.
2024 financial-services report
In December 2024, Treasury said it had received 103 comment letters in response to its request for information on AI in financial services. Its report recommended continued collaboration among government, regulators, and financial firms; examining regulatory gaps and potential consumer harm; coordinating risk-management frameworks; sharing AI-related information across the financial sector; and reviewing compliance before deployment. These were recommendations within the report’s scope, not binding rules. See Treasury’s report announcement.
2026 cybersecurity and risk-management initiative
In February 2026, Treasury announced a separate public-private initiative to develop practical AI cybersecurity and risk-management resources for financial institutions. Its workstreams include governance, data practices, transparency, fraud, and digital identity. Treasury described the effort as implementation resources; the announcement does not make those resources binding legal requirements. See Treasury’s initiative announcement.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the remarks establish—and what they do not
Bessent’s comments set out a policy stance: AI developers should own the work of managing risk, warnings should lead to solutions, and development should be accompanied by preparation and defense. They also describe a proposed international communication process, not a confirmed operating system.
Best Value
The figures and reports cited here do not measure how often AI incidents occur or predict the probability of cyber or biological harm. Treasury’s count of 103 letters is a count of submissions to its request for information, not a measure of AI risk. The interview reporting also does not independently verify a rate of incidents or provide a technical threshold for pausing model development.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
