We’re nearing the end of the month, which means it’s time for another installment of Brafton’s AI Roundup and Rundown series.
This month, updates span a spectrum of topics, from federal AI policy and affordability to tacky trends and some smaller good-to-knows from Google and the like that could impact the marketing industry in various ways.
Enjoy April’s updates!
A New AI-Focused Executive Order from the White House
The future of AI is still uncertain, but the U.S. is trying to stay ahead with a new executive order. Issued April 23, 2025, “Advancing Artificial Intelligence Education for American Youth” outlines a strategy to integrate AI literacy into the U.S. K-12 education system. Its main goals are to:
- Enhance AI understanding among Americans by integrating AI into educational curricula.
- Provide AI training for educators to effectively teach and utilize AI technologies.
- Introduce AI concepts and technologies early in the educational journey to develop an AI-ready workforce.
A Nigerian study from January showed just how effective AI can be in children’s education. Still, some folks are skeptical about the idea of integrating AI into early learning. How do you feel about it?
AI Is (or Was) Becoming More Affordable, Says Stanford University
Not only is AI becoming increasingly capable, but the cost of building and running models was (more on this in the next section) on a downward trend, according to Stanford University’s 2025 AI Index report.
The data suggests that “the inference cost for a system performing at the level of GPT-3.5 dropped over 280-fold between November 2022 and October 2024.”
AI leaders like OpenAI — and their users — are already beginning to experience the benefits of these lower costs. OpenAI recently launched Flex, a new API model designed to offer more affordable access to its AI models by trading off speed and availability. Flex processing is tailored for those lower-priority, non-production tasks such as model evaluations, data enrichment and asynchronous workloads, offering a 50% reduction in API usage costs.
Given what’s covered in the next section, we’ll see if AI companies can weather the storm and keep driving costs down, but I’m skeptical.
Tariffs Slowing Down AI Advancement in the U.S.?
Reuters released an article in the penultimate week of April discussing how escalating U.S.-China trade tensions and global economic uncertainties are impacting the AI sector. One of the major setbacks discussed is how tariffs are disrupting supply chains vital for AI infrastructure, particularly affecting the availability and cost of essential data center equipment, often manufactured in China. The cost of procuring necessary hardware for AI development has also risen. As a result, companies are delaying data center commitments or backing out of planned lease deals altogether, and it’s unclear when things will begin to bounce back.
Google Brought AI Overviews To 9 New EU Countries
At the tail end of March 2025, Google announced that users in the following countries will now have access to AI Overviews in Search:
- Austria.
- Belgium.
- Germany.
- Ireland.
- Italy.
- Poland.
- Portugal.
- Spain.
- Switzerland.
Despite some public hesitancy early on, Google continues to expand and improve its AI Overviews feature. As more countries are added to the list — and more marketers and SEO experts get to interact with the feature — perhaps we’ll learn new insights about how to leverage AIOs for SEO.
Apple Is Training AI Models on Synthetic Emails
Apple has championed user privacy for a long time, and they’re still holding strong to that tenet. While Meta and other leading AI companies have trained their models on real user-generated content, Apple announced that it’s using synthetic, AI-generated emails to smarten its algorithms.
There is still an element of realness here, though. Apple employs differential privacy techniques to compare the synthetic data with anonymized embeddings derived from real emails of users who have opted into the Device Analytics program. This method ensures that individual user content remains confidential while allowing the AI to learn from realistic patterns.
While this approach is ideal for balancing user privacy with the need for advanced AI functionalities, it introduces some risk, too, like potential biases or inaccuracies. That said, those risks are prevalent in human-created content as well. Time will tell how this one turns out for Apple and its algorithms!
‘Tacky’ GenAI Trends: What Should You Make of Them?
If you’ve been on social media the last few weeks, you’ve probably come across a couple of AI trends: the infamous Studio Ghibli trend that happened at the end of March, and the newer AI action figure trend from mid-April. These were so hot (and controversial) that I can’t not mention them here.
If you’re unaware, the Studio Ghibli trend saw millions of people feeding real images of themselves, loved ones, pets, etc. to ChatGPT in return for an impressive Studio-Ghibli-style render. It was met with major backlash. Many, many people were upset to see Hayao Miyazaki’s — co-founder of the animation studio — style ripped off by AI, especially since he’s spoken out about the technology and how he’s “utterly disgusted” by it — and that was 8 years ago.
The action figure trend was similar, only this time, users were generating playful representations of themselves and their lifestyles as toy action figures. As far as I saw, this trend wasn’t nearly as despised as the other, but it still had plenty of nay-sayers.
But why even mention these trends here? Well, because they were a pulse check of sorts on the public’s opinion on generative AI. The backlash folks saw from sharing images made clear one thing: It matters how you use these tools.
Social media users didn’t let regular people posting Ghilbi-esque images get away guilt-free, so you best believe they’d come down even harder on brands for similar wrongdoing. So, these trends serve as an important reminder to use GenAI as a tool, to use it intentionally and to always champion human artists and creativity over algorithmic outputs. It may be becoming increasingly easy to overlook in our industry, since it’s so fast-paced, but it’s an integral rule that brands will have to continue abiding by with intention to keep consumers on their side.
Wrapping Up
There’s a lot to digest here — some big changes that set a precedent for our future using AI, as well as some smaller items that are just good-to-know if you’re a marketer, like AI Overviews being available in more countries.
