Amazon's weekly operations meeting today reportedly focused on recent service outages and on the role that code changes attributed to generative AI may have played. However, the company is downplaying the possibility of problems with AI.
According to a briefing note for the meeting seen by the Financial Times, Amazon said that there had been a "trend of incidents" in recent months, characterized by a "high blast radius" and "Gen-AI assisted changes."
The implication is that AI-assisted coding has made the company's infrastructure more fragile. And that's something Amazon disputes.
Last Thursday, Amazon's website and ecommerce services were down for several hours for some users, an issue the company attributed to "a software code deployment." AWS was not involved, we're told.
The Financial Times report follows coverage last month that AWS's Kiro AI tool made system changes that affected the availability of AWS Cost Explorer in the Mainland China partition. Amazon, at the time, said that its AI tool was not to blame: "This brief event was the result of user error – specifically misconfigured access controls – not AI."
In the wake of that February incident, an Amazon spokesperson told The Register that "While security incidents involving misconfigured access controls can occur with any developer tool – AI-powered or not – we have not seen compelling evidence that incidents are more common with AI tools."
The Register asked an Amazon spokesperson whether, in light of the Financial Times' current claims, the company maintains that there's no internal evidence "that incidents are more common with AI tools." We were reassured that the statement stands, though the company has provided no data that would allow an independent analysis of incident causes.
An Amazon spokesperson said, "TWiST is our regular weekly operations meeting with a specific group of retail technology leaders and teams where we review operational performance across our store. As part of normal business, the meeting will include a review of the availability of our website and app as we focus on continual improvement."
Nonetheless, Amazon is extraordinarily attentive to this particular issue. Just five minutes elapsed between The Register's 0657 PT email inquiry and the 0702 PT call we received from a spokesperson during a dog walk.
Even for Amazon – which in recent years has responded assertively and rapidly when presented with challenging claims, unlike competitors that may not even respond – that pace of corrective messaging is surprising.
Writing for The Register last month, Corey Quinn, chief cloud economist at Duckbill, expressed disbelief at the company's assertion that AI was not to blame in the February outage. "AWS would rather have the world believe their engineers are incompetent than admit their artificial intelligence made a mistake," he said.
Other industry observers have raised similar concerns, arguing that headcount reductions compound the issues raised by AI.
Following the major AWS outage last October, James Gosling, the lead designer of Java, who left his role as distinguished engineer at AWS in 2024, said in a LinkedIn post that the company's focus on revenue generation at the expense of everything else resulted in layoffs to teams that didn't directly generate revenue but were still important for infrastructure stability.
"Back when the AI hype explosion happened and I was still at AWS I was astonished by how the structure of the business got torqued around, and how teams got demolished," Gosling wrote.
"The ROI analysis was disastrously shortsighted. These systems are complex interconnected structures. Unless the whole ecosystem is comprehended in total, bad decisions are made."
Following Gosling's departure, there were more layoffs, notably the 14,000 job cuts announced last October, justified by claims about the transformative nature of AI. ®
Source: The register