Sponsored by
Hey friends 👋
Big one today. The US government just ended its three-week standoff with Anthropic over Mythos, and it might have accidentally invented a whole new template for how frontier models get released. Also, Amazon is throwing $1 billion at "forward deployed engineers," which sounds great until you realize whose team they're actually on.
Let's dig in.
The government just paused frontier AI, and nobody planned it this way
AI safety folks have been asking for a pause on frontier model releases for years, and it always died the same way, because no lab wanted to go first and hand a competitor the advantage.
Well, it just happened anyway, and not because Anthropic or OpenAI chose it. Washington forced it.
The US Commerce Department slapped export controls on Anthropic's Mythos and Fable back on June 12, making it illegal for foreign nationals anywhere to access the models. The stated worry was that Mythos is remarkably good at finding software vulnerabilities, and nobody wanted that capability landing in the hands of countries like Iran, North Korea, Russia, or China. Those controls were lifted this week, and in a letter to Anthropic, Commerce Secretary Howard Lutnick said the company agreed to proactively flag security risks and keep the government looped in on future model releases.
OpenAI got similar treatment. GPT-5.6 was held to a limited preview instead of the wide launch OpenAI wanted, even though it reportedly benchmarks above Mythos in some areas, and Altman called the delay "bad news" on X.
Why this matters: whether you see this as government overreach or exactly the oversight AI needs, the outcome is the same. For three weeks, two of the most powerful models on Earth were held back by regulators instead of the labs that built them, and that has never happened before. If this turns into a real pre-release review process rather than a one-off scramble, it could reshape how frontier AI gets shipped going forward.
How AI-Era Pricing Is Reshaping Finance Operations
Usage-based and hybrid pricing models are changing how B2B companies generate revenue — and creating new headaches for the finance teams behind them.
Tabs co-founder Rebecca Schwartz and PwC Partner Amit Dhir sat down to unpack exactly what that means in practice: how pricing model decisions ripple into revenue recognition, forecasting, and financial ops — and what it takes to scale without piling on manual work.
Watch the on-demand recording to get practical frameworks, real-world examples, and a clear path to operationalizing usage-based revenue — including a forward-looking take on how AI will reshape financial workflows. If your team is navigating pricing complexity heading into the back half of the year, this is worth an hour.
Amazon's $1B forward-deployed-engineer bet has a catch nobody's saying out loud
AWS announced this week that it's putting a billion dollars behind embedding engineers directly inside customer companies to help them build and ship AI agents. The pitch is simple: get customers "self-sufficient with AI" quickly, and leave them with documentation, trained staff, and a repeatable playbook.
It seems to be working, since job postings for forward deployed engineers are up around 700% year over year, and OpenAI is running the same play through its Deployment Company.
Here's the catch, though. These engineers don't work for you. They work for Amazon, or OpenAI, or whoever is footing the bill. They're genuinely useful for lowering the barrier to enterprise AI, but by design they're also building your systems around their platform, and that's not a side effect. It's the business model.
Why this matters: if your company is considering bringing in forward deployed engineers, go in with your eyes open. You're not just buying expert help. You're also buying a fast, expensive path toward locking yourself into one vendor's ecosystem.
Hampton took $440K in planned hires off the calendar
Hampton co-founder Joe Speiser had three roles budgeted: a data engineer, an ops manager, a PM. $440K. He installed Viktor on April 12. Forty-four days later, none are on the calendar, and 18 of his team work with Viktor daily. His VP: we are editors now, not creators.
A few more things worth knowing
Apple's Doug Brooks gave an exclusive interview on how the company's decade-old bets on the Neural Engine, unified memory, and power efficiency are paying off in the agentic AI era, with Mac minis and Mac Studios reportedly seeing strong demand as local AI workhorses.
Chris Canal, CEO of eval vendor Equistamp, compared this moment to nuclear-style oversight, arguing that if AI is as impactful as labs claim, it probably needs the same kind of audits.
The open question nobody has answered yet is whether the Mythos and GPT-5.6 episode becomes a repeatable review framework or was just a one-time political fire drill.
That's what stood out to me today. Reply and tell me what caught your eye, I read everything.
Talk tomorrow,
Hatman 🎩