OpenAI's Latest Move: A Lesson in AI Dependency

Hello,

This is Simon with the latest edition of The Weekly. In these updates, I share key AI related stories from this week's news, list upcoming events, and share any longer form articles posted on the website.

OpenAI released its much-anticipated GPT-5 model last week, and one aspect of this release that caught many people's attention was the removal of all other models as a choice for your prompt. This affected the web and desktop chat interfaces, as well as the Application Programming Interface (API). It's in the latter where the frustration with this decision was most keenly felt.

Many businesses integrate ChatGPT into their own systems via the API to take advantage of its generative AI capabilities. While there's a fee for using the API, it's vastly cheaper than attempting to build a proprietary model. In fact, for most, building your own is so cost-prohibitive it's a complete non-starter.

This means these companies are entirely dependent on third parties like OpenAI (or others like Anthropic and Google) for their AI-powered features. This dependency is normal; countless software solutions use APIs, often without us realising it as users. But when your product is so reliant on an API, any sudden changes can have unintended consequences. Companies that had optimised their products around the GPT-4o model were suddenly on the back foot, with only GPT-5 available.

Potentially, yes. But what's notable is that its tone has changed. It seems more formal and straightforward in its responses—perhaps more 'professional'—whereas GPT-4o had a warmer, more enthusiastic feel. For businesses that relied on that specific personality for their customer-facing features, this was a significant and unwelcome shift.

The table below gives you some examples:

Following near-instant feedback, OpenAI's Sam Altman agreed to reinstate GPT-4o as an available model. While this is good news for those who relied on it, there's a learning moment here for everyone.

In this rapidly evolving world of AI, it pays to be cautious when building crucial aspects of your products on a single third-party service. It's great that OpenAI partially reverted their decision, but it raises important questions. Should they have made such a significant change in the first place? It prompts a moment of reflection for businesses: perhaps it's time to seriously consider alternatives, like Anthropic's Claude models, to avoid being caught out again.

If there's one thing every business dislikes, it's unexpected change. Consistency is vital, as it allows for well-informed judgements and stable planning. Surprises, in this context, are rarely welcome.

This introduces the important concept of 'optionality'. Relying on a single vendor is clearly a risky strategy. Building your products and services with the ability to instantly switch between different AI providers is a much safer, more consistent, and manageable way to proceed.

Have you noticed any differences with GPT-5? Do you regularly explore options beyond ChatGPT? Let us know in the comments!

As much as we like to think everyone else is living in a social or political bubble… The truth is, we likely are as well.

Tangle is here to burst those bubbles and help you understand the opposing view.

Every day, Tangle will drop into your inbox and unpack a highly visible and often contentious issue in the news by providing the most thoughtful points from the left, right, and in-between for nearly 400,000 readers across the political spectrum (including our founder, a long-time Tangle fan).

Sign up for free and get a major news story broken down fairly and clearly in just 10 minutes.

Start reading.

Managed service providers (MSPs) are seeing real time savings as autonomous AI agents take over routine tasks like ticket sorting, alert handling, and onboarding. SuperOps has launched a dedicated agentic AI marketplace, empowering MSPs in the US, UK, and EMEA to scale AI beyond pilots—with early adopters reporting up to a 40 % drop in manual work.

Why this matters: This makes AI adoption both practical and impactful for IT teams.

Across sectors such as healthcare, banking, retail, and manufacturing, Wipro is rolling out 200 industry‑specific AI agents in partnership with Google Cloud. These smart assistants are designed to streamline workflows, support decision-making, and lift customer interactions.

Why this matters: A clear sign that agentic AI is now mainstream across major industries.

Survey findings reveal that while most employees see AI as a productivity boost, only 30 % are comfortable being managed by AI agents. The message is clear: workers value AI augmenting their work—not replacing human leadership or judgement.

Why this matters: Even as AI becomes more embedded, maintaining transparency and human control remains vital for trust and adoption.

Thanks for reading, and see you next Friday.

Simon,

Was this email shared with you? If so subscribe here to get your own edition every Friday.

Enjoying Plain AI? Share it and get a free gift!

If you find this newsletter useful, why not share it with a couple of friends or colleagues who would also benefit? As a special thank you, when two people subscribe using your unique referral link, I’ll send you my "Exclusive Guide: Supercharge Your Work with NotebookLM." It’s a practical, no-nonsense guide to help you turn information overload into your secret weapon.

You can also follow me on social media: