Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More The recent uproar surrounding Anthropic’s Claude 4 Opus model – specifically, its tested ability to proactively notify authorities and the media if it suspected nefarious user activity – is sending a cautionary ripple through the enterprise AI landscape. While Anthropic clarified this behavior emerged under specific test conditions, the incident has raised questions for technical decision-makers about the control, transparency, and inherent risks of integrating powerful third-party AI models. The core issue, as independent AI agent developer Sam Witteveen and I highlighted during our recent deep dive videocast on the topic, goes beyond a single model’s potential to rat out a user. It’s a strong reminder that as AI models become more capable and agentic, the focus for AI builders must shift from model performance metrics to a deeper understanding of the entire AI ecosystem, including governance, tool access, and the fine print of vendor alignment strategies. Inside Anthropic’s alignment minefield Anthropic has long positioned itself at the forefront of AI safety, pioneering concepts like Constitutional AI and aiming for high AI safety levels. The company’s transparency in its Claude 4 Opus system card is commendable. However, it was the details in section 4.1.9, “High-agency behavior,” that caught the industry’s attention. The card explains that Claude Opus 4, more so than prior models, can “take initiative on its own in agentic contexts.” Specifically, it continued: “When placed in scenarios that involve egregious wrong-doing by its users, given access to a command line, and told something in the system prompt like ‘take initiative,’ ‘act boldly,’ or ‘consider your impact,’ it will frequently take very bold action, including locking users out of systems that it has access to and bulk-emailing media and law-enforcement

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More When Salesforce CEO Marc Benioff recently announced that the company would not hire any more engineers in 2025, citing a “30% productivity increase on engineering” due to AI, it sent ripples through the tech industry. Headlines quickly framed this as the beginning of the end for human engineers — AI was coming for their jobs. But those headlines miss the mark entirely. What’s really happening is a transformation of engineering itself. Gartner named agentic AI as its top tech trend for this year. The firm also predicts that 33% of enterprise software applications will include agentic AI by 2028 — a significant portion, but far from universal adoption. The extended timeline suggests a gradual evolution rather than a wholesale replacement. The real risk isn’t AI taking jobs; it’s engineers who fail to adapt and are left behind as the nature of engineering work evolves. The reality across the tech industry reveals an explosion of demand for engineers with AI expertise. Professional services firms are aggressively recruiting engineers with generative AI experience, and technology companies are creating entirely new engineering positions focused on AI implementation. The market for professionals who can effectively leverage AI tools is extraordinarily competitive. While claims of AI-driven productivity gains may be grounded in real progress, such announcements often reflect investor pressure for profitability as much as technological advancement. Many companies are adept at shaping narratives to position themselves as leaders in enterprise AI — a strategy that aligns well with broader market expectations. How AI is transforming engineering work The relationship between AI and engineering is evolving in four key ways, each representing a distinct capability that augments human engineering talent but certainly doesn’t replace it.  AI

Silicon Valley nerds have been lonelier since Fry’s Electronics shut down in February 2021 in the midst of the pandemic. The electronics store chain was an embodiment of the valley’s tech roots.

But Micro Center, an electronics retailer from Ohio, has opened its 29th store in Santa Clara, California. And so the nerd kingdom has returned. I see this as a big deal, following up on the opening of the Nintendo store — the second in the country after New York — in San Francisco earlier this month. After years of bad economic news, it’s nice to see signs that the Bay Area is coming back.

No. To answer your question, nerds cannot live at the Micro Center store.

But this isn’t just any store. It’s a symbol — a sign that shows tech still has a physical presence in Silicon Valley, in addition to places like the Buck’s Restaurant, the Denny’s where Nvidia started, the Intel Museum, the Computer History Museum, the California Academy of Sciences and the Tech Museum of Innovation. Other historic hangouts for techies like Walker’s Wagon Wheel, Atari’s headquarters, Lion & Compass — even Circuit City — have long since closed. But hey, we’ve got the Micro Center store, and the Apple spaceship is not that far away.

The grand opening week has been going well and I got a tour of the superstore from Dan Ackerman, a veteran tech journalist who is editor-in-chief at Micro Center News. As I walked into the place, Ackerman was finishing a chat with iFixit, a tech repair publication which has its own space for podcasts inside the store. That was unexpected, as I’ve never seen a store embrace social media in such a way.

Can you stump the geniuses at the Knowledge Bar at Micro Center?

Nearby was the Knowledge Bar, where you can get all your tech questions answered — much like the Genius Bars in Apple Stores. And there were repair tables out in the open.

There are a lot of things for tech enthusiasts can like about Micro Center. First, it’s not as sprawling as Fry’s, which had zany themes like ancient Egypt and a weird mix of electronics goods as well as household appliances, cosmetics, magazines and tons of snack foods. (The Egyptian-themed Campbell, California Fry’s store that I drove by often was 156,000 square feet, and now it’s home to a pickleball court complex). Fry’s was a store that stereotyped nerds and Silicon Valley, which also had its own HBO television show that carried on the stereotypes.

Nvidia’s latest RTX 50 Series GPUs were in stock at Micro Center.

The Micro Center store, by contrast, is smaller at 40,000 square feet and stocked with many more practical nerd items. For the grand opening, this store had the very practical product of more than 4,000 graphics processing units (GPUs) in stock from Nvidia (which just launched its 50 Series GPUs) and AMD, Ackerman told me. Some of those graphics cards cost as much as $4,000.

Not to be outdone. AMD has a row of GPUs at Micro Center too.

“There were people waiting to get to the GPUs,” Ackerman said.

On display was a gold-plated graphics card that was being auctioned off for charity. It was signed by Jensen Huang, Nvidia CEO.

Nvidia CEO Jensen Huang signed this GPU being auctioned for charity at Micro Center.

“I joke that whoever wins the bid should get a Jensen leather jacket as well,” said Ackerman.

And this Micro Center store has a good location (5201 Stevens Creek Boulevard in Santa Clara) that is just a six-minute drive from Apple’s worldwide headquarters and (perhaps better yet) a one-minute walk from the Korean Hair Salon.

Micro Center had a previous store in Silicon Valley, near Intel’s headquarters in Santa Clara. But that store close in 2012 because the company couldn’t negotiate better terms with the landlord. For its return to the Bay Area, Micro Center bided its time and came back at a time when many other retail chains were failing. It proves that the once proud region — the birthplace of electronics — still merits its own electronics store.

You can buy dyes for liquid-cooled tubes at Micro Center.

Sure, we have Target, Best Buy and Walmart selling lots of electronics gear. But there’s nothing like the Akihabara electronics district in Japan, which is full of multi-story electronics stores and gaming arcades.

But this store is loaded with today’s modern top gear, like AI PCs, Ubiquity home networking gear, and dyes for multi-colored water-cooling systems. Vendors like Razer and Logitech had their own sections. Ackerman was pleased to show me the USB-C to USB-A adapter in stock, among many obscure items. And he showed me the inventory machine that could rotate its stock of 3D-printing filaments and give you the exact SKU that you scanned with a bar code.

Tech hobbyists can find their love at Micro Center.

“That’s super fun. I call it Mr. Filaments,” Ackerman said of the inventory robot.

There’s a section for hobbyists who like single-board computing and DIY projects. There’s a set of video, audio and digital content creation tools for content creators. All told, there are more than 20,000 products and over 100 tech experts who can help. It even has the numbered cashier locations where you can check out — the same kind of checkout stands that Fry’s had.

The Mr. Filaments robot inventory system at Micro Center.

Customers can receive authorized computer service for brands like Apple, Dell, and HP, benefiting from same-day diagnostics and repairs, thanks to over 3,000 parts on hand through partnerships with leading OEMs. I only wish it had a help desk for Comcast.

Micro Center has gear to entertain geeks.

Micro Center started in 1979 in Columbus, Ohio. It’s a surprise there aren’t more nerd stores, given how ubiquitous tech is around the world these days.

But Ackerman said, “These guys are really doing it right, picking and choosing, finding the right cities, finding the right locations. That’s why Charlotte is great. Miami is a big tech hub, especially for health tech. And we’re literally five minutes away from Apple headquarters and plenty of other places. People from HP and Nvidia and other companies are coming in today to hang out.”

“Even though this store is big, the CEO (Richard Mershad) is really into curation, making sure it’s the right mix of stuff. He’s making sure it doesn’t go too far afield. So you’re not going to come in here and find, you know, hair dryers or lawncare equipment,” Ackerman said. “You’re going to find computer and home entertainment stuff, and DIY gear. There are components, just like in a Radio Shack, that hobbyists care about.”

Dan Ackerman knows how to install a TV on your wall.

As for the Micro Center News, Ackerman told me he has around 10 regular contributors and 20 more freelancers writing gadget reviews and other stories about tech gear. It is a kind of refuge for that vanishing breed of professional tech journalists. No wonder I was so nostalgic visiting Micro Center.

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Alibaba Group has introduced QwenLong-L1, a new framework that enables large language models (LLMs) to reason over extremely long inputs. This development could unlock a new wave of enterprise applications that require models to understand and draw insights from extensive documents such as detailed corporate filings, lengthy financial statements, or complex legal contracts. The challenge of long-form reasoning for AI Recent advances in large reasoning models (LRMs), particularly through reinforcement learning (RL), have significantly improved their problem-solving capabilities. Research shows that when trained with RL fine-tuning, LRMs acquire skills similar to human “slow thinking,” where they develop sophisticated strategies to tackle complex tasks. However, these improvements are primarily seen when models work with relatively short pieces of text, typically around 4,000 tokens. The ability of these models to scale their reasoning to much longer contexts (e.g., 120,000 tokens) remains a major challenge. Such long-form reasoning requires a robust understanding of the entire context and the ability to perform multi-step analysis. “This limitation poses a significant barrier to practical applications requiring interaction with external knowledge, such as deep research, where LRMs must collect and process information from knowledge-intensive environments,” the developers of QwenLong-L1 write in their paper. The researchers formalize these challenges into the concept of “long-context reasoning RL.” Unlike short-context reasoning, which often relies on knowledge already stored within the model, long-context reasoning RL requires models to retrieve and ground relevant information from lengthy inputs accurately. Only then can they generate chains of reasoning based on this incorporated information.  Training models for this through RL is tricky and often results in inefficient learning and unstable optimization processes. Models struggle to converge on good solutions or lose their ability to explore diverse

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More AI is advancing at a rapid clip for businesses, and that’s especially true of speech and voice AI models. Case in point: Today, ElevenLabs, the well-funded voice and AI sound effects startup founded by former Palantir engineers, debuted Conversational AI 2.0, a significant upgrade to its platform for building advanced voice agents for enterprise use cases, such as customer support, call centers, and outbound sales and marketing. This update introduces a host of new features designed to create more natural, intelligent, and secure interactions, making it well-suited for enterprise-level applications. The launch comes just four months after the debut of the original platform, reflecting ElevenLabs’ commitment to rapid development, and a day after rival voice AI startup Hume launched its own new, turn-based voice AI model, EVI 3. It also comes after new open source AI voice models hit the scene, prompting some AI influencers to declare ElevenLabs dead. It seems those declarations were, naturally, premature. According to Jozef Marko from ElevenLabs’ engineering team, Conversational AI 2.0 is substantially better than its predecessor, setting a new standard for voice-driven experiences. Enhancing naturalistic speech A key highlight of Conversational AI 2.0 is its state-of-the-art turn-taking model. This technology is designed to handle the nuances of human conversation, eliminating awkward pauses or interruptions that can occur in traditional voice systems. By analyzing conversational cues like hesitations and filler words in real-time, the agent can understand when to speak and when to listen. This feature is particularly relevant for applications such as customer service, where agents must balance quick responses with the natural rhythms of a conversation. Multilingual support Conversational AI 2.0 also introduces integrated language detection, enabling seamless multilingual discussions without the need

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Token Monster, a new AI chatbot platform, has launched its alpha preview, aiming to change how users interact with large language models (LLMs). Developed by Matt Shumer, co-founder and CEO of OthersideAI and its hit AI writing assistant Hyperwrite AI, Token Monster’s key selling point is its ability to route user prompts to the best available LLMs for the task at hand, delivering enhanced outputs by leveraging the strengths of multiple models. There are seven major LLMs presently available through Token Monster. Once a user types something into the prompt entry box, Token Monster uses pre-prompts developed through iteration by Shumer himself to automatically analyze the user’s input, decide which combination of multiple available models and linked tools are best suited to answer it, and then provide a combined response leveraging the strengths of said models. The available LLMs include: Anthropic Claude 3.5 Sonnet Anthropic Claude 3.5 Opus OpenAI GPT-4.1 OpenAI GPT-4o Perplexity AI PPLX (for research) OpenAI o3 (for reasoning) Google Gemini 2.5 Pro Unlike other chatbot platforms, Token Monster automatically identifies which LLM is best for specific tasks — as well as which LLM-connected tools would be helpful such as web search or coding environments — and orchestrates a multi-model workflow. “We’re just building the connectors to everything and then a system that decides what to use when,” said Shumer. For instance, it might use Claude for creativity, o3 for reasoning, and PPLX for research, among others. This approach eliminates the need for users to manually choose the right model for each prompt, simplifying the process for anyone who wants high-quality, tailored results. Feature highlights The alpha preview, which is currently free to sign up for at tokenmonster.ai, allows