Hello everyone! I’m excited to share the latest developments in AI, robotics, coding, and more in this week’s edition of TekTonic. This week brings stories from GPT -5’s roadmap to fresh insights on AI’s impact on business, cybersecurity, and beyond. Let’s dive into the headlines together, and afterward, don’t miss my short LinkedIn and Twitter posts summarizing these updates. Happy reading!
This week’s notable stories didn’t make the news headlines but rather the research feeds. Three papers show remarkable conclusions about the AIs we are creating. First, researchers discovered language models develop a coherent value system as they scale, and those value systems converge with other language models’ value systems. They also found that, as LLMs scale, they oppose changing their values. The second major research news comes from Berkley, where they confirmed creating thinking models from language models is cheap and easy. Those researchers took a model 1000 times smaller than OpenAI’s flagship and beat it using the Deepseek R1 reinforcement learning method. Finally, a separate group unveiled a language model capable of “silent” reasoning—similar to how humans use mental imagery or unspoken logic. In this setup, the model’s thinking loop happens in hidden layers instead of tokens, enabling stronger problem-solving and greater versatility than traditional approaches. All three discoveries promise to shape upcoming AI models in the months ahead.
Now, on to the news. Sam Altman stood firm this week and declared that OpenAI is “not for sale,” following Elon Musk’s unsolicited $97.4 billion offer. In another development, Altman teased GPT -5’s upcoming release—promising a unifying AI system incorporating search, voice, and advanced reasoning. Tensions between Altman and Musk continued when Altman commented on Musk’s lifelong “insecurities,” sparking debates about the personal dynamics driving AI competition. Meanwhile, Microsoft published a study warning that excessive reliance on AI at work could erode critical thinking skills, adding another dimension to the ongoing conversation about AI’s societal impact.
A creative security loophole emerged when a software engineer harnessed Unicode quirks to embed concealed data in emoji characters, highlighting new possibilities—and risks—in everyday communication. Meanwhile, scientists at Caltech developed the first programming language for “living matter,” pushing the boundaries of synthetic biology and showing how code might soon govern cellular behavior and material properties.
The business landscape felt more AI-related tremors as Meta moved forward with laying off 5% of its workforce, coinciding with rising IT unemployment attributed partly to AI-driven automation. At the same time, an executive statement from Microsoft emphasized how advanced AI capabilities are ready to reshape banking and investment services.
AI’s influence on media and entertainment was highlighted in several ways. AMD revealed discussions with game developers about using AI to generate immersive virtual worlds, potentially revolutionizing game design. Meanwhile, OpenAI’s Super Bowl commercial for ChatGPT underscored that human creativity still has a role in marketing AI’s power, as humans made the ad entirely. Finally, Bay Area artists urged Christie’s Auction House to reconsider an AI art auction, expressing concerns over unauthorized data usage and the future of creative labor.
New research from Anthropic suggests that coding queries are the most common use case for AI chatbots, reinforcing how developers leverage AI as a coding copilot. Reflecting this trend, OpenAI unveiled “Large Reasoning Models” for competitive programming, where their o3 model scored a gold medal in the International Olympiad in Informatics (IOI).
A reported “breach” of OpenAI made headlines in the security arena, but researchers clarified that infostealer malware, not a direct hack, was the culprit. This finding highlights the persistent threat of password-harvesting tools and the importance of robust endpoint security.
Nvidia’s newest GPUs again drew scrutiny after reports of melting power connectors on the RTX 5090 surfaced, echoing previous controversies. Meanwhile, OpenAI accelerated efforts to finalize its first custom AI chip design—targeting less reliance on third parties—and Meta explored acquiring Korean startup FuriosaAI to bolster its chip capabilities in the race for AI hardware supremacy.
Microsoft expanded the capabilities of its Edge Copilot by adding a new share button, allowing users to distribute AI chat sessions easily. Additionally, Perplexity AI launched its “Deep Research Alpha,” offering in-depth web analysis features, advanced reasoning, and convenient PDF export—aiming to rival major players in the advanced AI research assistant space.
At the University of Alabama, a research team is teaching students to code by letting them pilot drones with their minds. This novel approach uses specialized brain-computer interface software, giving K-12 learners a thrilling, hands-on experience that sparks fresh enthusiasm for programming.
In government news, Anthropic has partnered with the UK’s Department for Science, Innovation and Technology to explore how AI can improve public services. Their collaboration includes real-world trials of Claude, Anthropic’s AI model, and analysis of AI-driven labor market shifts, aiming to guide policy decisions responsibly.
Robotics made notable strides as Boston Dynamics enhanced its humanoid robot, Atlas, in partnership with the Robotics & AI Institute, focusing on advanced locomotion and manipulation. Meanwhile, Google joined a $350 million funding round for Apptronik, a company developing humanoid machines that aim to compete with Tesla’s Optimus—raising the stakes in the race for humanoid robot innovation.
Anthropic, a prominent AI startup, is projecting potential revenue of up to $34.5 billion by 2027, reflecting the explosive growth potential of generative AI services and the surging demand for advanced language models—especially if it continues to close large funding deals and ramp up technology development.
The 2024 Evident AI Index provided a fresh look at how banks across North America, Europe, and APAC rank in AI maturity—a helpful benchmark tool for any organization building data-driven services. Finally, rumors swirled around Apple’s future hardware endeavors, as analyst Ming-Chi Kuo predicted Apple might start building robots by 2028, signaling a potential leap forward in the company’s evolving smart home ecosystem.
That is it. Stay tuned for more news next week. I APPRECIATE ALL THE READERS. Help spread the word. I put these newsletters together so productive humans like you can stay on top of the latest AI stories shaping our world.
If you like these weekly tech news reports, subscribe to get notified of new editions and updates. For daily updates, check out our news page. For a more in-depth analysis of the week’s news, sign up for our free weekly newsletter to the right of the daily news, or follow me on Twitter or YouTube.
