The AI Systems Running Your Company May Soon Need Their Own Passports
AI is growing up fast — and today’s news shows the awkward growing pains that come with it. A startup just raised $66 million to solve a problem most people haven’t thought about yet: what happens when an AI has access to your company’s systems and nobody knows how to verify it’s supposed to be there? Meanwhile, the White House locked down one of Anthropic’s most advanced models after signs it may have reached Chinese hands. And several major AI companies are quietly lining up to go public. These stories aren’t separate. They’re all symptoms of the same shift: AI is moving from a tool we use to a system we have to govern.
AI Agents Are Getting Hired. Now Someone Has to Check Their IDs.
AI agents — software programs that can take actions on their own, like booking meetings, sending emails, or querying databases — are showing up in workplaces at scale. The problem is that most companies have no reliable way to know which agents are authorized to do what. NewCore, a startup founded by people who previously worked in cybersecurity and crypto, just raised $66 million to fix that.
Think of it like building a badge system for an office. Humans get keycards that say who they are and what rooms they can enter. AI agents, right now, often have no equivalent. They can slip into systems using borrowed credentials or vague permissions, and nobody notices until something goes wrong. NewCore wants to create a formal identity layer for these agents — a way to say, with confidence, “this AI is authorized, this one isn’t, and here’s exactly what it’s allowed to touch.”
For regular employees and managers, this matters in a practical way. If your company starts using AI agents to handle invoices, customer data, or internal communications, you’d want to know those agents can’t be hijacked or spoofed by a bad actor. As TechCrunch reports, NewCore is betting that identity management for AI will become as essential as antivirus software once was.
Why this matters: Every company adopting AI agents is quietly building a security liability they don’t fully understand yet. NewCore is early, but the problem it’s solving is real and growing fast.
“NewCore emerges with $66M to rebuild identity security for the agentic era.”
The White House Just Restricted an Anthropic AI Model Over China Concerns
Export restrictions — government rules that control which technologies can be shared with other countries — aren’t new. But applying them to a specific AI model is a significant escalation. According to The Verge, the White House moved to restrict Anthropic’s Mythos model after concerns that a Chinese-linked group may have accessed it without authorization.
The trigger, reportedly, was a warning from Amazon’s CEO. Amazon has a deep financial relationship with Anthropic, so an alarm from that level carried weight. The fear isn’t just that someone looked at the model. The concern is that a sophisticated enough actor could study how it works, replicate its capabilities, or find ways to exploit it. Advanced AI models represent years of expensive research, and the gap between leading and trailing systems matters strategically.
For people outside the tech industry, this is a signal worth paying attention to. Governments are starting to treat advanced AI the same way they treat fighter jet technology or missile guidance systems — as something you don’t let leave the country freely. That framing will shape regulation, investment, and which AI tools your company can legally use depending on where you operate.
Why this matters: AI security is no longer just a corporate IT problem. It’s becoming a foreign policy issue, and the rules are being written right now.
“White House export restrictions on Anthropic’s Mythos linked to Chinese access concerns.”
Several Major AI Companies Are Eyeing the Stock Market
Initial public offerings — when private companies sell shares to the public for the first time — are notoriously hard to time well. According to TechCrunch, Anthropic, OpenAI, and others are reportedly exploring IPOs, hoping to ride current enthusiasm for AI before market conditions shift. Specific valuations and timelines remain unclear, so treat these as early signals rather than firm plans.
Why this matters: If these companies go public, their incentives change. Shareholders want profits, not just research. That pressure will influence which AI products get built and how quickly.
Also Happening in AI
OpenAI had a busy day beyond the IPO rumors. The company announced it’s acquiring Ona, a startup that handles cloud infrastructure for AI agents, suggesting OpenAI wants more control over the technical plumbing its agents run on. It also launched new OpenAI Academy courses to help workers use AI tools more effectively on the job. Elsewhere, Google quietly redesigned its search box for the first time in 25 years, according to VentureBeat, making room for longer, more conversational queries — a direct response to how people now talk to AI. And TechCrunch reports that an Earth observation satellite successfully identified a target on its own, without ground instructions, pointing toward a future where AI-powered hardware makes decisions in real time, far from human oversight.
What to Watch
The NewCore funding and the Mythos restrictions are connected threads. As AI agents gain access to sensitive systems, the question of who controls them — and who can steal them — becomes urgent. Watch for more governments to announce AI-specific export rules in the next 60 days, and for enterprise security companies to start positioning AI identity as their next major product category. The companies that figure out AI governance early will have a real advantage over those scrambling to catch up.