ENTERPRISE

IBM just gave OpenAI a shortcut into the enterprise

As forward deployed engineers become the hottest new role in AI, OpenAI is trying forward deployed partners.

On Thursday, the AI lab announced a partnership with IBM to embed its technology into IBM's consulting arm. Through the partnership, IBM's forward-deployed units, trained through the OpenAI Partner Network, will help enterprise clients modernize their workflows with OpenAI's tech. 

Additionally, IBM will join OpenAI's "elite partner tier," and will launch a dedicated OpenAI Practice that includes thousands of consultants and engineers, specifically trained with OpenAI certifications. 

Colleen Kapase, OpenAI's vice president of global partnerships and ecosystem, exclusively told The Deep View that partnering with IBM represents a "significant" opportunity. "IBM works with enterprises around the world and across virtually every major industry, including many of the complex and highly regulated environments where AI can have enormous impact."

The partnership has three areas of focus: 

• Updating legacy operations with OpenAI's frontier models, including GPT-5.6 and products like Codex and ChatGPT Work, across domains including finance, procurement, customer operations, and HR

• Modernizing legacy applications and software development using Codex and ChatGPT Work, aiming to help enterprises simplify engineering processes and speed up product shipment

• And increasing IBM's participation in Project Daybreak by combining OpenAI's cyber capabilities with IBM Autonomous Security, its agentic security platform, to help clients manage both cyber and AI model risk

The goal of the partnership is to help IBM Consulting enterprise clients get beyond one of the major hurdles standing in the way of successful AI deployments: The fact that so much enterprise work relies on fragmented legacy systems and tools that aren't well fitted for the modern and fast-paced nature of AI, Andy Baldwin, global senior vice president of IBM Consulting, told The Deep View. 

"Most organizations don’t struggle with getting access to AI tools or models anymore. The real challenge is putting AI to work securely at scale and effectively across large, complex organizations," said Baldwin. "At the same time, AI is changing the threat landscape. Organizations are facing more sophisticated, AI-enabled threats and growing pressure to strengthen and build agentic first cyber defenses."

This partnership comes as enterprises face a particularly significant crossroads. Many enterprises want to embed AI throughout their organizations. However, as models get stronger, the cybersecurity risks grow more daunting, with models from Anthropic, OpenAI and Meta all revealing powerful cybersecurity capabilities (and accidental breaches) that surprised even the researchers creating and testing them. As such, enterprises walk a fine line between constantly adopting while rapidly securing. On top of that, many organizations are facing mounting pressure to see returns as costs grow. A partnership like this could help organizations strike the balance between all three factors, or it could lock organizations into using OpenAI products instead of potentially cheaper alternatives, such as open and custom models. 

TOGETHER WITH CRUSOE

GLM-5.2 - Fine-tune and deploy your own instance

GLM-5.2 is Z.ai’s flagship long-horizon coding model, featuring a usable 1M-token context window that reliably maintains API contracts and engineering intent where others fail. Stop managing GPU clusters.

With Crusoe Serverless Fine-Tuning, you customize GLM-5.2 on your proprietary data in a tenant-isolated environment—no data sharing, no infrastructure overhead. When your job finishes, deploy to Self-Serve Deployments in one click or download raw .safetensors weights for full portability. Every run is reproducible, auditable, and fully yours. No lock-in. No guesswork. Experience production-grade, customized performance without the DevOps burden.

Start Fine-Tuning GLM-5.2 now

PRODUCTS

Why Pixel 11 uses AI to simply make a better phone

AI has gotten so divisive that one of the world's leading tech brands is in no hurry to label all of its leading innovations as AI features.

On Wednesday, Google officially took the wraps off its new line of Pixel devices at its Made by Google event in New York, hosted by comedian Trevor Noah. Among the phones, watches, foldables, and earbuds announced, the flagship device was the new Pixel 11 family that bolstered its reputation as the world's best AI-first smartphone. 

The Pixel 11 series has packed in new AI features that are very promising, based on my first impressions from the demos in New York as well as my first 24 hours with a review unit. Google isn't shying away from AI marketing hyperbole, saying Android is no longer an operating system but an intelligence system. Still, it made the adroit move of framing many of the best new features on their own merits. 

"I think the word [AI] is actually a little scary," Sameer Samat, Google's president of Android ecosystem, told The Deep View, "and I think it kind of turns some folks off. So one of the things that we’re really trying to do is just talk about the benefits of the features… even if [we] never mention the word AI."

So far, these are the three AI-powered features that stand out the most:  

1. Rambler voice keyboard: If you've used Wispr Flow, then you know how much faster it lets you input text than a keyboard and how much more accurate and easier to use it is than the traditional dictation on Siri or Google Assistant. But as good as Wispr Flow is on a computer, it's very clunky to use on both iPhone and Android. By contrast, Rambler is built directly into Google Keyboard (Gboard) and so it's as fast and as easy to access as the old dictation feature, but works so much more effectively. 

2. Magic Capture photos: The sleeper feature in Pixel 11 is the new Magic Capture mode. It's now a third capture mode alongside photo and video capture. When you switch into this mode, you can record a scene and then Magic Capture will automatically produce a video and use its algorithm to produce several still images that are the best high-resolution options (not screen grabs or low-quality frames). This is especially impressive with moving objects where it can freeze motion and pick the most dramatic pose. 

3. Multi-step tasks: You can now type or talk to your Pixel, describe a goal or series of actions you'd like it to take and it can carry out those tasks using an AI agent in a safe, sandboxed environment that you can monitor and that will prompt you for approval when it needs to carry out the final actions such as purchasing something, sending a message, and/or scheduling time on your calendar. I appreciate it that Google is slow-walking this one to help the broader public get comfortable with AI agents.

There are a lot more AI-powered features on the Pixel 11, but from a long-term perspective, the agentic features are the most revolutionary and disruptive. That's why Google is both moving deliberately in rolling out agentic powers to its latest phones, while also shifting its marketing for Android to center the future of the platform around its agentic mission. "It's no longer an operating system that helps you use your phone," said Rick Oserloh, Google's SVP of devices, during the New York launch event. "It's an intelligence system that gets things done for you." Make no mistake, that's going to change the way you use your phone significantly in the long run. Today's phones are still focused on apps. Once the agentic paradigm sets in, tomorrow's phones will be more centered around tasks, goals, and actions. See my full interview with Android lead Sameer Samat on The Deep View Conversations for more on how that future is unfolding.

TOGETHER WITH JUMPCLOUD

Meet the Zombie Agents already inside your environment

An agent gets deployed. The project ends. The developer moves on. The agent keeps running, keeps accessing systems, and keeps accumulating permissions with no owner, no review, and no offboarding process.

Research from JumpCloud found that only 21% of organizations have governance controls in place for non-human identities. Zombie Agents are not a future risk.

Find out what securing every identity actually looks like →

PRODUCTS

Why Mozilla is on a mission to prevent AI lock-in

Mozilla's Firefox once helped shape the web as we know it. Now, the veteran browser is setting its sights on tech's next frontier.

When Anthony Enzor-DeMeo became CEO of the organization in December 2025, he made clear his intention to integrate AI into the browser. However, he is taking a unique approach: giving users the ability to choose which model they want in the browser's AI chatbot sidebar. Model choice is also a primary feature of its upcoming launch of Smart Window, AI built into Firefox that aims to prioritize privacy. 

In a conversation with The Deep View, Enzor-DeMeo explained how Mozilla is "late to the party" on AI by design, using that delay to understand consumer sentiment and differentiate on model choice rather than building its own models. 

This interview has been edited for brevity and clarity.

Sabrina Ortiz: Why was now the right time for Mozilla to pivot toward AI?

Anthony Enzor-DeMeo: We are developing Smart Window, and that's been in development almost a year, so it's probably more accurate to say we're late to the party. But for a few reasons: One I think it's been interesting to see what other players have done; and two, I think it takes time to understand consumer sentiment.

Sabrina Ortiz: You have decided to give people the choice to use their own models. Was that because you saw demand from your audience, or did you just feel that was the best approach?

Anthony Enzor-DeMeo: It's a mixture of the two. The interesting thing about a browser is that almost everyone on the planet uses it, so you're appealing to many different audiences. The more technical audience will understand what the plan is [and] what the community [wants], those types of things. A more general audience might say, "I want something that's fast. I want something that's secure." So we've had to do a lot of testing to find the right word choices that would incentivize people to choose X or Y or Z. 

Sabrina Ortiz: If you’re not building frontier models, is there any interest there? If not, what gives Mozilla its unique value?

Anthony Enzor-DeMeo: I have no plans to develop a frontier model. We are owned by a nonprofit. I don't think it's the best use of our capital. We obviously have capital, but I think it could be deployed elsewhere. We're going to be focused much more on promoting the ecosystem, and so we want to obviously promote open source, open weight. 

Sabrina Ortiz: We are seeing a huge emphasis on open-weight and open-source. Why is this a priority for you?

Anthony Enzor-DeMeo: You can learn a lot from looking at the past. What open-source did for the internet, 20-plus years ago, was create like $8 trillion worth of value, according to an HBS study that came out. That's largely because once you standardize and get people all working on the same infrastructure, the same code, it becomes easier to innovate. What I want to see is more innovation fostered. I want to see more competition, and one way to do that is to give access. 

And the interesting thing about those who choose to give access is if you're giving something away, it's obviously easier to gain adoption, and then more people will conform to those standards, and thus those people will be in more control, and we saw that with the web. I think it's important that the US and globally everyone's supporting more open-source and open-weight models because we want to foster that innovation. We don't just want one or two players doing it because then you'll run into a very similar situation: if they're just doing it and everyone uses the same free ones, then suddenly there are no standards.

LINKS

• AI coding startup Cognition in talks to raise $1 billion at $40 billion valuation

• Mistral will now support third-party open models, starting with Z.ai

• Bank of America commits $250 billion to US infrastructure, data centers

• Charles Yang joins Anthropic Institute as chief of staff

• UK government to regulate AI use for gene synthesis, bioweapons

• Vibe coding firm Lovable raises $400 million at $13 billion valuation

• Grok 4.6: SpaceXAI's latest model, with improved performance on long-running agents and more "ambitious" interactive and visual work

• Claude Cowork: Anthropic's work agent is now available in the Chrome side panel, allowing it to take on tasks in your browser and work across tabs.

• Wan-Animate-2: The latest animation model from Alibaba, featuring high-fidelity character animation and real-time streaming generation. 

• ChatGPT Work and Codex: Users can now keep work from other agents in sync with those from OpenAI products by importing projects, chats, skills, and plugins.

• OpenAI: Model Designer

• Sphere: AI Agent Infrastructure Lead

• Adobe: Operations AI Developer

• Waymo: AI Enablement Lead

GAMES

POLL RESULTS

Do you think you're usually able to distinguish authentic from AI-generated content?

Yes (13%)
Somewhat (66%)
No (19%)
Other (2%)