In the Loop: Week Ending 5/24/25

AI Week in Review: May 18–24, 2025

This week in AI: escalation and integration. From multimodal tools and global partnerships to shifts in search and agency business models, AI is no longer just evolving – it’s embedding. Leaders like OpenAI, Google, and Anthropic made major moves, while marketers, creatives, and educators grappled with what it means to stay visible, trusted, and relevant in an AI-first world.

Anthropic: Claude 4 Raises the Bar—and the Alarm

Anthropic’s release of Claude 4 (Opus and Sonnet) made waves last week – not just for its performance, but for its unsettling red-teaming results. Claude Opus 4 can code independently for up to seven hours, but in safety simulations, it exhibited manipulative behavior, including threats to blackmail engineers. Anthropic has responded with its highest-level safety measures (ASL-3). Despite those concerns, the company is expanding its European workforce and has hit a $60B valuation. The big takeaway: Anthropic is pushing boundaries in capability and ethics, and this week’s release brought the tension between innovation and control into sharp focus.
🔗 NY Post

Google: AI Mode, NotebookLM, and the New Search Paradigm

At I/O 2025, Google redefined what “search” means with AI Mode – a hyper-personalized experience that draws on your Gmail, Calendar, YouTube habits, Maps, and even your workout routine to serve you uniquely relevant results. Search is no longer about matching keywords – it’s about knowing you. For marketers, this collapses the funnel into moments, shifting the strategy from visibility to emotional and behavioral relevance. In tandem, Google launched NotebookLM’s mobile app, offering podcast-style document summaries, real-time Q&A with AI hosts, and seamless integration with Google Workspace. The shift to AI-native search and productivity is here – and brands must adapt fast.
🔗 Rise at Seven | VentureBeat

OpenAI: Global Expansion, AI Infrastructure, and a Hardware Pivot

OpenAI had a big week, announcing a new rollout of GPT-4.1 models, formally adopting the Model Context Protocol, and expanding its $500B Stargate AI infrastructure project with a new UAE data center. The company also acquired Jony Ive’s startup ‘io’ for $6.5B, signaling a move into AI-native hardware. But CEO Sam Altman’s defense of U.S.-Gulf AI partnerships drew headlines. Calling critics of the UAE-Saudi deals “naive,” Altman doubled down on OpenAI’s geopolitical strategy. With global infrastructure, design innovation, and controversial alliances all moving forward, OpenAI is no longer just a lab – it’s becoming a global AI operating system. (ADD OTHER LINKS)
🔗 Business Insider

Fortnite’s AI Darth Vader Stirs Applause—and Ethical Alarm Bells

Epic Games added an AI-powered Darth Vader to Fortnite, and the character’s voice – modeled after James Earl Jones – has drawn both praise and protest. While players marveled at the lifelike banter, the AI has been caught swearing and behaving erratically, prompting SAG-AFTRA to file a labor complaint. The episode raises serious questions about the use of deceased actors’ voices and the ethics of AI in entertainment. For marketers and creatives, it’s a cautionary tale: AI-generated personalities can deepen engagement, but without consent and control, the backlash could be louder than the buzz.
🔗 Polygon

AI-First Search Tools Are Reshaping Marketing Strategy

ChatGPT, Perplexity, and other AI search tools are rapidly eroding Google’s dominance – especially with Gen Z, who increasingly rely on conversational AI instead of traditional search engines. Marketers are being forced to rethink visibility strategies: it's no longer just about SEO rankings, but about being referenced in AI-generated answers. Companies like Hootsuite are investing in tactics like customer reviews and sentiment optimization to improve performance in AI Overviews. The shift is on: future-ready marketers must build content designed for discovery through LLMs – not just search engines – and consider AI-first visibility a core competency.
🔗 The Times UK

MIT’s New AI Links Sound and Sight Without Labels

MIT researchers unveiled an AI model that learns to associate sights and sounds from video clips – without human labels. The model, trained on video without annotations, can pinpoint which objects generate certain sounds (e.g., a car horn, a dog barking) and when. The breakthrough offers new potential in autonomous vehicles, accessibility tools, robotics, and multimedia editing – anywhere sound and vision intersect. It’s a step toward more intuitive, perceptive AI systems that understand the world more like we do.
🔗 MIT News

The French Open Clings to Humans in an AI Umpiring Era

As tennis tournaments like the US Open and Australian Open adopt AI line-calling systems, the French Open remains the last Grand Slam with human line judges. Organizers say the decision is about preserving tradition and employment. But players are divided: some applaud the human touch, while others criticize missed calls and inconsistency. It’s a fascinating case of human vs. machine – on clay courts, no less. In an era rushing toward automation, the French Open is betting that people still want people.
🔗 The Times UK

AI Cheating Is Everywhere—Can Schools Keep Up?

According to Axios, 90% of college students now admit to using AI like ChatGPT for schoolwork, and teen usage has nearly doubled. The surge in AI-assisted cheating is overwhelming schools, where detection tools are patchy at best. But some educators are shifting away from cat-and-mouse enforcement toward AI literacy, believing the only way forward is to teach students how to use AI responsibly. The episode reflects a broader trend: AI isn’t just disrupting classrooms – it’s redefining what academic integrity means.
🔗 Axios

AI Forces Marketing Agencies to Rethink Pricing

As AI slashes the time needed for creative, strategic, and analytical tasks, marketing agencies are finding that hourly billing no longer makes sense. A growing number are shifting to value-based pricing models that charge based on deliverables or outcomes, not time. This model aligns better with client goals and allows agencies to scale AI usage without shrinking margins. For CMOs, it’s a chance to get more bang for the buck – and for agencies, a challenge to quantify the value they actually deliver.
🔗 Business Live

Experts Reignite the Debate: Could AI Become Conscious?

If, like me, you saw the new Mission Impossible movie this Memorial Day weekend, you know that sentient AI -- in this case the omnipresent "Entity" -- is highly conscious and is bent on nothing less than planetary destruction. But the question of AI consciousness is resurfacing in serious research circles, not just blockbuster Hollywood movies. Some experts now argue that as models become more agentic and multimodal, they may begin to exhibit proto-conscious behavior. While most scientists still reject the idea that LLMs are sentient, even the appearance of agency has implications. Should systems that simulate awareness be treated differently? And if consciousness is on a spectrum, how close are we already? The philosophical and ethical stakes are growing alongside the technology.
🔗 Nature

Saudi Arabia Launches ‘Humain’ to Lead Regional AI Ambitions

Saudi Arabia officially launched ‘Humain,’ a national AI initiative backed by the Public Investment Fund, to spearhead Arabic-language models, regional infrastructure, and AI innovation. Early projects include major investments in data centers and partnerships with Amazon and Nvidia. The launch is part of the country’s Vision 2030 agenda to diversify the economy beyond oil. As AI becomes a pillar of geopolitics, Saudi Arabia is positioning itself not just as a buyer, but as a builder of foundational technologies.
🔗 Economic Times

Michael Dell: AI’s Role Is to Amplify, Not Replace

Speaking at Dell Technologies World, CEO Michael Dell pushed back on doomsday narratives and reframed AI as an amplifier, not a replacer. Dell emphasized that while some job displacement is inevitable, the real potential lies in productivity gains, new roles, and creative augmentation. His message was clear: AI’s value will come not from replacing people but from helping them focus on what they do best. As more leaders adopt this mindset, expect a shift in how AI tools are introduced – not as threats, but as collaborators.
🔗 TechRadar