GOVERNANCE

OpenAI expands Mythos rival to wider audience

AI-powered cybersecurity is poised for a major surge.

On Tuesday, OpenAI announced that it was scaling up its Trusted Access for Cyber (TAC) program to thousands of “verified individual defenders” and hundreds of teams involved in protecting critical software. These users will have access to a GPT-5.4 variant called GPT-5.4-Cyber that was trained to be “cyber-permissive”. 

Specifically, the model was fine-tuned to have fewer capability restrictions and includes new capabilities for these workflows, such as binary reverse engineering, according to the post. In the coming months, OpenAI says it will fine-tune its models to better cater to cybersecurity needs. 

Because of the permission updates to GPT-5.4-Cyber, OpenAI is first deploying it to vetted security vendors, organizations, and researchers who are part of the TAC's new tiers of access, with people willing to work with OpenAI to authenticate themselves as cybersecurity defenders. Interested TAC members apply for the cyber defenders pilot via this Google form. 

TAC users will still be able to take advantage of the benefits introduced with the program in February, including reduced friction with safeguards on models. Becoming part of the TAC is meant to be simple: individuals apply with the form, and enterprises request trusted access for their team through their OpenAI rep.

OpenAI also re-emphasized its mission and strategy for cybersecurity-related AI solutions, with key highlights including the need to expand access while preventing misuse, to deploy iteratively, and to invest in the ecosystem. 

This announcement, of course, follows Anthropic’s launch of Project Glasswing, an initiative to secure critical software in the age of AI, which included the launch of Claude Mythos, a model more advanced than its current top-tier model, Opus. The biggest difference is that Anthropic’s launch was much more limited, with only launch partners and 40 additional organizations getting access. 

“Anthropic has taken a more contained route, with a clear focus on how these systems behave when pushed beyond their intended use,” said Hanah-Marie Darley, chief AI officer and co-founder of Geordie AI, an AI Governance platform. “OpenAI is opening access more broadly and placing weight on the safeguards around the model. Both approaches are reasonable.”

Anthropic's Mythos announcement dropped while I was at HumanX last week, making it a hot topic in nearly every conversation I had there. A common thread was that withholding the model was a calculated move for media coverage, investor interest, and safety-first branding. I disagree. The decision still reflects a sense of responsibility, because Anthropic had to know that competitors like OpenAI had similar models and would release them more broadly, given their different risk tolerance. Choosing restraint meant ceding ground and facilitating competition. As for OpenAI’s approach, expanding access also inevitably entails greater risk. 

TOGETHER WITH TIGER DATA

agents are desperate for ephemeral databases

they spin up projects, fork environments, test ideas, and tear them down. over and over. but every database on the market was designed for humans who provision once and stick around. agents don't work that way.

ghost is a database built for agents. unlimited databases, unlimited forks, 1 TB of storage, and 100 compute hours per month. all free. try it here.

PRODUCTS

Gemini desktop app signals tighter AI race

It took a while, but Google AI has arrived on the desktop app scene. Here is what it brings to the table, and whether it was worth the wait.

On Wednesday, Google unveiled the Gemini app on macOS as a native desktop experience. Ultimately, it is meant to make it easier for users to access Gemini’s assistance, no matter what they are viewing on their screen, through simple shortcuts. 

If you have ever used the Claude or OpenAI macOS apps, the concept is essentially the same: share your screen with Gemini to get help with whatever you are looking at, including local files. The extent of what you can get assistance with or ask about is the same as that of interacting with the new Gemini app. 

Ultimately, Google is positioning this as a way to eliminate the need for context switching to increase workflow speed and productivity: 

“Switching between windows on your desktop can be clunky and slow. Now, you can bring up

Gemini from anywhere on your Mac with a quick shortcut (Option + Space) to get help instantly,

without ever switching tabs,” said Michael Friedman, Group Product Manager, Gemini App, in the post. 

While there, context switching is indeed one of the biggest productivity bottlenecks for working professionals. I did get deja vu writing this article, reminiscent of the ChatGPT desktop app articles I wrote back in 2024 when it launched. But like they say, better late than never. 

All you need to do is download the app at no extra cost to get started. It is available to macOS users on Sequoia (15.0) or later and runs exclusively on Apple Silicon. There's no sign of a Windows version of the desktop Gemini app yet. But OpenAI and Anthropic also both launched on Mac first and then followed up with a Windows version. Google said it will have more news to share in the coming months, which could be a nod toward a Windows version.

While Google is now one of the most formidable players in the AI race, it was not always that way. The launch of Google Bard was rocky, and it wasn’t until the rebranding to Gemini that the company began to show promise, so much so that now it has one of the best LLMs on the market, and even Apple has now chosen it to power Siri. However, the release of this app highlights that in some ways, Google is still playing catch-up. Bridging those gaps is more necessary than ever with Claude Code and Cowork becoming agentic and growing in popularity. And, of course, OpenAI is reportedly working on a “superapp” that combines ChatGPT, Atlas, and Codex. 

TOGETHER WITH MERGE

How teams plan to use MCP this year

Most teams building AI agents plan to adopt the Model Context Protocol (MCP) this year. Most of those same teams have serious security concerns about it.

To understand how teams are navigating this tradeoff, we surveyed hundreds of AI leaders building AI agents for the first-ever state of agentic integrations report.

Their top concerns?

• 70% worry about credential leaks and malicious servers

• 56% say MCP doesn’t support enterprise search well

• 51% report ambiguous tool definitions causing incorrect tool calls

Get your free copy to learn more.

SECURITY

Why enterprises need agent-era security now

What happens when AI agents outnumber humans in the enterprise?

In this episode of The Deep View Conversations, Senior Reporter Sabrina Ortiz sits down with Jeetu Patel, President and Chief Product Officer at Cisco, to explore how the rise of AI agents is reshaping cybersecurity, software development, and enterprise strategy.

Jeetu makes a bold prediction: today’s 150 million developers could expand to 3 billion agent builders within the next year. But that explosion comes with serious risk. As nation-states and bad actors deploy autonomous agents at scale, traditional, human-centered security models begin to break down.

This conversation unpacks:
+ Why AI agents are becoming the new attack surface
+ What enterprises must do now to prepare for agent-driven threats
+ Jeetu’s journey from Box to Cisco, and what it taught him about leading through platform shifts
+ Practical advice for learning AI and building in an agent-first world

But this isn't a doom-and-gloom conversation. Jeetu lays out a vision for how security can become an accelerator rather than a limiter, and why the distinction between giving agents access and giving them trusted, governed access will define which enterprises thrive in the agentic era.

If AI agents are the next platform shift, cybersecurity may be the defining battleground.

📺 Watch on YouTube

🎧 Listen in your favorite podcast player

Subscribe for more interviews with the leaders shaping the future of AI.

LINKS

• Allbirds, the shoe company, pivots into an AI compute provider company

• Apple threatens to remove Grok from App Store due to sexualized deepfakes

• Wired analysis found 600 students globally impacted by AI deepfake nudes

• Meta, Broadcam expand partnership to develop next-gen custom silicon  

• AI supercloud startup Parasail raised a $32 million in Series A

• Andreessen, Horowitz give $25 million to pro-AI super PAC

• Granola: We stopped taking notes in meetings and started getting better ones. Granola captures everything while you stay focused on the conversation. Try it free with code THEDEEPVIEW. (sponsored)

• Gemini 3.1 Flash TTS: Google’s “most controllable & expressive text-to-speech model”

• Midjourney V8.1: Offers 3x faster + cheaper native 2K HD rendering than V8 

• Google Chrome: Skills in Chrome lets users save and easily call on AI prompts 

• ElevenLabs: Seedance 2.0 is live in ElevenCreative

• OpenAI: Threat Modeler, Preparedness

• Fireworks AI: Member of Technical Staff, Research

• Nvidia: Senior Research Scientist, Fundamental Generative AI

• RAND: Senior Research Lead - AI Security Portfolio

GAMES

A QUICK POLL BEFORE YOU GO