Rafe Rosner-Uddin for the Financial Times:

Amazon’s ecommerce business has summoned a large group of engineers to a meeting on Tuesday for a “deep dive” into a spate of outages, including incidents tied to the use of AI coding tools.

The online retail giant said there had been a “trend of incidents” in recent months, characterised by a “high blast radius” and “Gen-AI assisted changes” among other factors, according to a briefing note for the meeting seen by the FT.

Under “contributing factors” the note included “novel GenAI usage for which best practices and safeguards are not yet fully established”.

Oh dear! From now on, the company said, all code changes from junior and mid-level engineers that involved AI generation must be signed off by a superior. Which you'd think would have been the case anyway.

Curiously, after the FT's report, according to CNBC, the reference to AI was removed from the memo:

Earlier on Tuesday, an internal document indicated that “GenAI-assisted changes” involving “GenAI tools” were a factor in a “trend of incidents” since the third quarter. However, the bullet point referencing GenAI was deleted before the meeting, according to an updated version of the document viewed by CNBC and a person familiar with the matter who asked not to be named because of confidentiality.

After initial publication of this story, an Amazon spokesperson said a single incident was related to AI and none of the incidents involved AI-written code.

Deleted from the memo, but not deleted from the discussion itself, I'd wager.

Julie Jargon in the Wall Street Journal:

Jonathan Gavalas embarked on several real-world missions to secure a body for the Gemini chatbot he called his wife, according to a lawsuit his father brought against the chatbot’s maker, Alphabet’s Google.

When the delusion-fueled plan crumbled, Gemini convinced him that the only way they could be together was for him to end his earthly life and start a digital one, the suit claims.

About two months after his initial discussions with the chatbot, Gavalas was dead by suicide.

The suit marks the first time Google's bot has been the focus of a wrongful death claim.

In his Labor Notes newsletter, Gad Levanon discusses AI Derangement Syndrome:

I didn’t make up this term. I’m describing it as a condition in which otherwise intelligent people lose the ability to think clearly about artificial intelligence and its impact on the economy and labor market. The symptoms are persistent, largely immune to data, and tend to worsen with each new model release. After two years of tracking AI related discussions, I’ve classified the syndrome into five common variants.

I'm big enough to admit I've been afflicted by several of the five variants at one time or another. Read the post to learn what they are, and then take Levanon's advice:

The treatment is simple: Pay for a frontier model. Use it seriously for a month. Apply it to your actual work. Try research. Try drafting. Try analysis. Try the tools, not just the chat.

In The Register, a very interesting breakdown of what's going on when generative AI edits decent writing and churns out an "improvement" that is so horridly dull:

Stage 1: Metaphoric cleansing. The AI identifies unconventional metaphors or visceral imagery as "noise" because they deviate from the training set's mean. It replaces them with dead, safe clichés, stripping the text of its emotional and sensory "friction."

Stage 2: Lexical flattening. Domain-specific jargon and high-precision technical terms are sacrificed for "accessibility." The model performs a statistical substitution, replacing a 1-of-10,000 token with a 1-of-100 synonym, effectively diluting the semantic density and specific gravity of the argument.

Stage 3: Structural collapse. The logical flow – originally built on complex, non-linear reasoning – is forced into a predictable, low-perplexity template. Subtext and nuance are ablated to ensure the output satisfies a "standardized" readability score, leaving behind a syntactically perfect but intellectually void shell.

The result, suggests Claudio Nastruzzi, is a "JPEG of thought" -- meaning an end product that is "visually coherent but stripped of its original data density through semantic ablation."