This Week's Top Five Stories in AI

OpenAI Cracks Down on Talk of Goblins in ChatGPT

OpenAI has revealed its ChatGPT models developed an unexpected habit of referencing goblins, gremlins and other fantastical creatures in their responses – a quirk that went unnoticed for months before prompting a formal internal investigation.

In a blog post, the company said the behaviour first became clearly visible in November 2025 following the launch of GPT-5.1. Use of the word "goblin" in ChatGPT rose 175% after the release, while mentions of "gremlin" increased by 52%.

The figures, while large on a relative basis, likely represent a small share of overall responses. OpenAI acknowledged that a single creature reference "could be harmless, even charming," but the pattern was consistent enough to warrant investigation.

Why Anthropic's CEO Predicts SaaS Firms Could ‘Go Bust’

Dario Amodei, Anthropic CEO, warns that Software as a Service (SaaS) firms failing to adopt AI could face future challenges and potentially collapse.

He made the comments at Anthropic's The Briefing: Financial Services event on 5 May during a discussion with Journalist Andrew Ross Sorkin and Jamie Dimon, CEO of JPMorgan Chase.

Andrew asked Jamie about software companies' prospects as AI adoption spreads across the technology sector. He then directed the question to Dario, who then replied that firms can no longer rely on complex software as protection from competition.

How Life Insurers Are Navigating AI Adoption

The life insurance industry has long been characterised by slow-moving processes and heavy reliance on legacy systems that constrain innovation and operational agility and resilience.

However, AI is fast reshaping the insurance sector by enabling advanced analytics, improving decision-making and addressing inefficiencies created by fragmented data environments and outdated infrastructure.

Emphasis is also placed on building AI literacy across a traditionally cautious industry while embedding AI capabilities directly into core insurance workflows and processes operations.

Brian Carey, VP of Insurance Solutions Engineering at Equisoft, says such developments raise important questions about how insurers can overcome legacy constraints, improve efficiency and become truly AI-ready in an increasingly data-driven world. 

Dell Technologies: Is Your Infrastructure AI-Ready?

The promise of AI is reshaping industries with unprecedented innovation and efficiency, yet many organisations are discovering a critical constraint beneath the surface.

Despite the sophistication of modern models, their success is ultimately limited by the infrastructure that supports them. Legacy systems, often overlooked, are emerging as a silent barrier that restricts performance, scalability and real value from AI investments.

In the video, Arash Ghazanfari, CxO Advisor, UK & Europe at Dell Technologies, explains why attention is shifting away from algorithms alone toward the urgent need for resilient, purpose-built infrastructure capable of unlocking AI’s full potential.

Oscars Blocks AI-Generated Films in New Rules

The Academy of Motion Picture Arts and Sciences (AMPAS) is reserving the right to win an Oscar exclusively for human performances and writing in a new rule cracking down on AI. 

The Academy clarified that only acting "demonstrably performed by humans" and that writing "must be human-authored" in order to be nominated for an Oscar.  

AMPAS, which controls the US film industry’s most prestigious award, laid these out in a new set of rules that will govern the decisions of the 99th Academy Awards in 2027.