In wake of outage, Amazon calls upon senior engineers to address issues created by 'Gen-AI assisted changes,' report claims — recent 'high blast radius' incidents stir up changes for code approval

(Image credit: Getty / Caroline Brehman)

Amazon allegedly called its engineers to a meeting to discuss several recent incidents, with the briefing note saying that these had “high blast radius” and were related to “Gen-AI assisted changes.” According to the Financial Times, one of the contributing factors listed in the meeting notes was the use of generative AI tools “for which best practices and safeguards are not yet fully established.”

Go deeper with TH Premium: AI and data centers

Microsoft data center in Mount Pleasant, Wisconsin — (Image credit: Microsoft)

There has been a spate of problems in Amazon’s operations recently, including a six-hour disruption on its main retail website, wherein customers were unable to see details and complete transactions, which the company said is attributed to erroneous code deployment. We’ve also seen reports that Amazon’s AI assistant could be easily jailbroken to answer questions unrelated to shopping, as well as reports of AI coding bot-driven outages with AWS, the company’s cloud service.

“Folks, as you likely know, the availability of the site and related infrastructure has not been good recently,” Amazon Senior Vice President Dave Treadwell allegedly said in an email, according to the publication. He also said that the meeting will take a “deep dive into some of the issues that got us here as well as some short immediate term initiatives,” and that AI-assisted changes must now be approved by senior engineers before deployment. This meeting is reportedly usually optional, but it seems that Treadwell asked the staff to be in attendance this time.

Article continues below

Amazon hasn’t officially confirmed the cause of all its woes, but the details shared in this meeting seemingly point to the use of AI. “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," an Amazon spokesperson told Tom's Hardware. "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.” It also isn’t the first big tech company to take things seriously after many firms took the “move fast and break things” motto literally when it came to generative AI. Microsoft said in late January 2026 that it’s working to fix many of Windows 11's flaws and restore its reputation. This came nine months after CEO Satya Nadella said that AI writes up to 30% of the company’s code, with some projects completely coded by AI.

While generative AI does have its uses, especially in specialized fields like medical research, it still needs observation, and we still cannot rely on its output 100% of the time. Unfortunately, many are overselling the capabilities of this tool, and many CEOs aren’t getting the promised benefits of higher revenues and reduced costs.

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

TOPICS

Jowi Morales is a tech enthusiast with years of experience working in the industry. He’s been writing with several tech publications since 2021, where he’s been interested in tech hardware and consumer electronics.

3 Comments Comment from the forums

ravewulf

The real solution is simple: stop using Gen-AI
Reply
Pete Mitchell

The problem isn't Gen-AI itself, it's how it's being used. Companies like Microsoft have been selling it as the solution for everything, and brain-dead executives are buying into the hype. This is why companies looking to sell software and services love talking to high-level execs. They know if they talked to the people who actually worked with the technology they'd get laughed out of the office.
Reply
Al26

How can all of these businesses have forgotten that big success requires significant risk.
i.e. INVESTMENT.
Good business leaders are just really good at managing those risks on aggregate.
Do none of these businesses understand the concept of investment.
*AI is still very young. If it's not costing you substantially more to implement, at this stage, than to not implement, then you're not doing it right.*Those costs must include people, because AI is currently in, not just development, but 'research' and development phase, and that doesn't at all stop with the model builders themselves. That means your normal review and continual improvement routines just won't cut it. You need to have some separate, additional teams, who's dedicated function and expertise is standing outside normal operations, observing both people and tooling, and *directing* tasks, whose sole and only purpose and value, is to test operational processes, in 'experimental' fashion, who's express goal will often include explicitly to do so in a way that *must" not contribute to immediate business value, because it's sandboxed of from the live production business. Introduced into production processes only when it serves R&D needs, and pulled at moment when it conflicts over some threshold, with sandboxed live business priorities. For none of this am I talking about mere code either. Its about ways of working.
Sadly this applies to all businesses at all scales with run of the mill middle management who have never in their life experience ever handled the kind of 'experiment' that AI is.
The "right attitude", just isn't going to cut it. In fact that will only make their jobs, recognising the existence of problems unimagined, all the harder.
Its staggering complacency. When everyone has forgotten that AI is even a new thing, that is when you can start being complacent about it.
Or, just keep dumping all that risk (which is very unknown, ie high) on your core business interests.
Even if things are working now, are you confident you have a handle on even quantifying the levels or rate technical debt you might be accruing? I think technical debt is a concept which many non technical businesses fail too understand that they have their own perfect analogues of
Just how deep are the waters underneath you already?
Reply