Nadella expects more outages as AI-found bugs force rushed patches
Microsoft's CEO told the DealBook Summit that AI is finding flaws so fast that fixes go straight to production with little testing, adding instability.
Key takeaways
- Satya Nadella said AI uncovers software flaws more quickly than companies can safely repair them, and that he expects this to cause more outages.
- He said patches now go straight into production because flaws are exploited immediately, leaving little time for testing.
- Nadella apologized for Microsoft's recent Outlook outage and said it was not directly linked to the patching problem.
- He warned that much critical infrastructure runs firmware that cannot be patched.
- He said that if alignment does not catch up with scaling, "we are in a world of hurt."
Microsoft chief executive Satya Nadella said AI now uncovers software vulnerabilities more quickly than companies can safely repair them, and that this is likely to cause more outages, TNW and TRT World report. He spoke on Wednesday at the DealBook Summit held by The New York Times.
In the past, a patch for a critical banking system would be tested for about a week, Nadella said, according to TNW. Now, because attackers exploit flaws as soon as they surface, patches go straight into production. "The patch needs to go instantly from being available to being patched in production, because it's being exploited in the wild. So that's going to create its instability," he said, according to TNW. TRT World reports he warned that rushing out fixes with less testing time could introduce new bugs.
What did he say about Microsoft's own outage?
Nadella apologized for the recent Outlook outage and said it had no direct link to the patching problem, TNW reports. "This was a latent bug that was there. It got invoked and, you know, it's one of those perfect storms," he said. TRT World says the outage, on November 13, hit Microsoft 365 services around the world. Later that day, on CNBC's Mad Money, Nadella said Microsoft could have done more to prevent it, TNW reports: "Of course, we should have done the testing before."
How does he see the security risk?
Nadella said the "floor of cyberattacks has fallen," TNW reports. AI could enable "essentially a nation-state level persistent attacker going after you all the time, so that means we have a lot of fragility," he said, according to TRT World. He also said much critical infrastructure runs firmware that cannot be patched, naming chillers, heating and cooling systems, and transport networks, TNW reports. Asked by host Andrew Ross Sorkin whether this was a passing phase, he drew a comparison with Y2K but said he had not settled on a view, since models keep getting better.
What else did he discuss?
On AI safety, Nadella said scaling had delivered results, but alignment, and a reliable way to debug models, had lagged behind, TNW reports. "I mean, if it doesn't happen, we are in a world of hurt," he said. He called for what he described as a trust architecture, including tamper-proof logs of model reasoning that other models can inspect.
Nadella also spoke about Microsoft's offer of voluntary exit packages to about 60,000 employees, saying it was meant "to make sure that at least everyone felt like they had agency in making choices," TRT World reports. He said software engineering remains a strong profession but the way teams work is changing, citing an 11-person team behind Microsoft's Cowork agent.
According to TNW, Nadella said routine office tasks do not require frontier models anymore, and that Copilot already chooses among models from OpenAI, Anthropic and Microsoft's own MAI line. Microsoft holds stakes in both OpenAI and Anthropic, TNW notes.
Disclosure: NextFish's AI drafting desk uses Claude, a model made by Anthropic.
Questions readers are asking
Why does Nadella expect more outages?
He said AI tools find vulnerabilities so quickly that fixes must go into production at once because attackers exploit them, which leaves less time to test and can add instability.
Was Microsoft's Outlook outage caused by AI-found bugs?
Nadella said it was not: he said it had no direct link to that problem, TNW reports, and that a "perfect storm" of events triggered a previously undetected bug, according to TNW and TRT World.
Where did Nadella make these comments?
At the DealBook Summit run by The New York Times, in conversation with Andrew Ross Sorkin, and later on CNBC's Mad Money.
Sources
- #microsoft
- #satya-nadella
- #outages
- #cybersecurity
- #patching
- #dealbook
Get the day's top AI & tech stories on WhatsApp: Follow the NextFish channel
Get stories like this every morning.
Related
Policy & Safety2 min
Pentagon CIO Kirsten Davies named acting US cyber director
Trump said Friday he fired Sean Cairncross because he was not 'good enough'; other officials had described a resignation. The change follows a series of US tech outages.
Cybersecurity2 min
F5 issues emergency fixes for critical BIG-IP remote code execution flaws
The bugs can crash BIG-IP's traffic engine and may let unauthenticated attackers run code. F5 lists no mitigations besides the hotfixes.
Cybersecurity2 min
Critical NGINX flaws can let attackers run code on web servers
F5 has patched five remote code execution bugs. Wiz says AI tools helped it find the two HTTP/2 bugs; no attacks in the wild have been reported.
Cybersecurity3 min
Hijacked Microsoft Agent Framework package on PyPI hid a credential stealer
Two malware-laced releases of a Microsoft AI developer package were live on PyPI for under an hour. Researchers say anyone who installed them should rotate every exposed secret.