Nadella warns AI-driven bug discovery could lead to more outages
Microsoft’s chief executive said urgent security fixes leave less time for testing, but said the company’s Outlook outage was not directly linked to that problem.

Microsoft chief executive Satya Nadella said he expects service outages to become more frequent as artificial intelligence uncovers software vulnerabilities faster than companies can safely address them. Speaking at The New York Times’ DealBook Summit in New York, he warned that pressure to deploy security patches immediately could cause instability.
Nadella apologized for an outage that disrupted Outlook, while saying it was not directly connected to the risks he described. He attributed that incident to a previously hidden bug. In a separate interview with CNBC’s Jim Cramer, he acknowledged that Microsoft should have tested more and put safeguards in place to prevent a global disruption.
According to Nadella, a critical banking system’s patch once would have undergone a week of testing. Companies now face pressure to install fixes immediately because attackers are already exploiting the vulnerabilities. He also warned that critical infrastructure, including heating and cooling equipment and transport networks connected to the internet, uses firmware that cannot be patched.
On AI oversight, Nadella said greater model scale had not yet delivered alignment or a reliable way to debug models. He advocated prohibiting models whose internal reasoning people cannot read and requiring tamper-proof reasoning logs that other models could inspect. He said organizations deploying AI systems are currently liable for their actions and predicted insurers would seek those logs when assessing risk.




