Three separate stories broke this week that, taken together, sketch out where the AI industry is heading over the next couple of years. According to The Information’s latest TITV video roundup, OpenAI’s new Astra model has reignited the argument over how close we are to artificial general intelligence, Anthropic is moving into payments, and rising AI-driven threats are forcing companies to rewrite their security budgets. Each one matters on its own. Together, they tell you the frontier labs are done being research shops. They’re becoming platforms that want to sit inside your workflows, your wallet, and your risk models.
OpenAI reopens the AGI question
The Information reports that OpenAI’s Astra has pushed the AGI debate back to center stage. This is the pattern now: every major model release gets framed as a step toward general intelligence, and every claim draws an equal and opposite wave of skeptics. What stands out here isn’t whether Astra qualifies as AGI. It’s that the label itself has become a business tool. Calling something ‘close to AGI’ shapes funding rounds, recruiting, and the pricing power a lab can command.
For practitioners, the takeaway is to ignore the label and test the capability. Run Astra against your actual tasks. Benchmarks and marketing language tell you less than one afternoon of real evaluation on your own data.
Anthropic goes after payments
The more concrete signal is Anthropic pushing into payments, as detailed in The Information. This is a big deal because it changes what an AI company is. Handling transactions means an agent doesn’t just draft the email or plan the trip. It completes the purchase. That’s the leap from assistant to operator.
It also puts Anthropic on a collision course with a crowded field:
- Stripe and traditional payment rails, which agents will need to plug into or replace
- OpenAI, which has its own commerce ambitions inside ChatGPT
- Banks and card networks that own the trust layer today
Whoever owns agent-driven payments owns a cut of every transaction an AI completes on your behalf. That’s why a model company is suddenly interested in moving money. If you’re building agent products, start thinking now about how checkout, authorization, and spending limits work when a bot holds the card.
AI threats are moving security budgets
The third thread is the one most companies feel first: AI threats are shifting how businesses spend on security. The Information’s reporting lines up with what we’ve seen elsewhere, including recent cases of infostealer malware draining Claude and other AI accounts. Attackers now use AI to scale phishing, generate malware, and hijack the very AI tools teams rely on.
The budget response is already underway. Money is moving toward identity protection, credential monitoring, and controls built specifically for AI accounts and API keys. This is significant because it flips the ROI story. Companies bought AI to cut costs. Now they’re spending to defend the AI they bought.
What to do about it
The common thread across all three stories is that frontier labs are expanding out of the model and into the stack around it. Payments, security, and the AGI narrative are all fights over who controls the layer where AI meets real money and real risk.
A few practical moves for the next 12 to 24 months:
- Evaluate on capability, not claims. Treat ‘AGI’ as marketing. Test models against your workflows.
- Plan for agent payments. If your product touches commerce, design spending limits and authorization now, before agents start buying.
- Fund AI-account security. Rotate API keys, monitor credentials, and protect AI logins like you protect banking access.
- Watch the platform play. The lab that owns payments plus models plus security controls owns your dependency. Keep your options portable.
The labs are racing to become infrastructure, not just intelligence. The companies that plan for that shift now will negotiate from a stronger position when it arrives. Full details are in The Information’s original report.