We didn't see it coming. Or maybe we did — and just didn't want to admit it out loud.
Sam Blackshear, the engineer who created Move, the programming language meant to fix everything broken about smart contracts, just announced he's joining Anthropic to work on AI safety. Not as a part-time advisor. He's leaving Mysten Labs, the company he co-founded, the team behind the Sui network. The news hit the industry like a wave, and SUI holders immediately started doing mental math.
I've audited Move-based protocols. I've run workshops where developers' eyes widened when they realized a language could make reentrancy attacks almost impossible by design. I've watched, over the past year, as the AI industry quietly began eating our lunch. This isn't just another hiring memo. It's a signal about where the hardest technical problems — and the people who love solving them — are actually going.
Let's dig into what this really means.
The Man and the Language
Blackshear's resume reads like a security researcher's dream. At Meta (then Facebook), he worked on the Diem and Novi projects, where he designed Move from scratch. Move wasn't just another smart contract language. It was built around resource safety — the principle that digital assets can't be duplicated, dropped, or accidentally reused. Combined with formal verification, it made certain classes of hacks structurally impossible rather than merely inconvenient. In Move, the compiler itself rejects code that mishandles asset ownership. You don't need to be a security expert to avoid the most common exploits; the language just refuses to compile.
When regulators crushed Diem, Blackshear co-founded Mysten Labs in 2021. The team built Sui, an L1 blockchain that bet everything on Move's safety guarantees to stand out against Aptos, Solana, and Avalanche. In a market full of "fastest chain" marketing battles, Sui's pitch was different: "Our language is safer, so our DeFi is safer, so our ecosystem actually deserves your money." It was a message aimed directly at developers exhausted by a decade of drained treasury exploits and bridge hacks. And for a while, it worked beautifully.
That pitch worked. Sui forged a real community, attracted serious DeFi talent, and carved out a distinct identity. It built its entire narrative around a single person's creation.
Now the architect of that safety narrative is walking out the door.
What Blackshear Actually Brings to Anthropic
Here's the part most coverage gets wrong.
Blackshear is not a typical Web3 developer. He's a programming language designer with deep expertise in static analysis, type theory, and formal verification. Those skills don't map to "AI safety" as most people imagine it — red-teaming chatbots, writing RLHF pipelines, or debating alignment taxonomies on social media.
They map to something much more specific: proving that a system behaves exactly as intended under adversarial conditions.
That is precisely what AI safety lacks today.
In my own work bridging these fields, I've stared at this gap. We've spent ten years building tools to verify that a smart contract won't lose user funds. Model checkers. Proof assistants. A culture that repeats "don't trust, verify" until it's muscle memory. AI systems, by contrast, are enormous stochastic black boxes. Nobody can prove a frontier model won't leak training data, fall for a prompt injection, or make a disastrous decision under pressure. In blockchain, we take it for granted that a critical contract gets audited and formally verified before billions of dollars sit behind it. In AI, the equivalent practice remains largely aspirational. Regulators keep asking for safety standards, but standards without tools are just intentions.
Anthropic, which has branded itself the "safety-first" AI lab, isn't hiring Blackshear to write blog posts. They're hiring him to build. The formal verification mindset that emerged from blockchain's security battles might well become the backbone of AI safety auditing. The question of "is this AI system actually safe?" may soon be answered with machine-checkable proofs, not vibes.
We didn't invent verification, but we industrialized it. That experience is now exportable.
The Sui Problem Nobody Is Talking About
Now here's the uncomfortable part for the Sui community.
We didn't talk enough about the single-point-of-failure embedded in Sui's technical narrative. Move was the moat. And in the public imagination, Blackshear was Move.
The Move language didn't evolve on its own. It had a specific human champion inside Mysten Labs pushing compiler improvements, formal verification tooling, and language roadmap decisions. When that person leaves — even gracefully — the language's momentum shifts. Unless Mysten Labs promotes or hires a worthy successor quickly, Move's rate of innovation will slow. And in this market, standing still means falling behind.
To be fair: this is not "Sui is dead." Mysten Labs retains serious engineering depth. CEO Evan Cheng is a compiler and systems veteran from Apple and Meta, and the broader team includes cryptographers and systems engineers who can absolutely keep Sui running and evolving. But the public narrative gap is real. Sui needs a new "technical face" — quickly — or the ecosystem's story starts to feel like a house losing its architect.
But here's what genuinely worries me: the message this sends to every builder watching.
When a co-founder of a top-tier L1 walks over to an AI company, builders everywhere start doing the math. I've seen it happening in my own events. At hackathons this year, a growing number of developers are asking about AI infrastructure rather than DeFi protocols. The gravity well is shifting. And if compensation and perceived impact keep tilting toward AI, we will keep losing people like Blackshear.
The Contrarian Read: Maybe We Deployed Him
Now for the counterintuitive angle.
Maybe this isn't a loss for blockchain. Maybe it's the beginning of a cross-pollination that saves both industries.
Here's the thing about formal verification: it took years to convince blockchain projects that proving properties of smart contracts was worth the effort. Now, as AI systems grow more powerful and more dangerous, the exact same argument is about to be replayed on a much larger stage. The AI industry is going to need tools for verifying model behavior, auditing inference pipelines, and proving that systems won't do things they shouldn't.
And we've been building those tools for more than a decade. The Move community, the Rust verification folks, the ZK proof crowd — we've all been circling the same insight: trust should be mathematical, not anecdotal.
If Blackshear brings Move-inspired rigor to AI safety, that's a genuine win for humanity. And if he builds bridges back to crypto — AI-assisted smart contract auditing, provable inference chains, verifiable AI agents — then this "defection" transforms into something else entirely: a beachhead.
The deepest exchanges between these fields haven't happened yet. Blockchain's obsession with verifiability and AI's need for trustworthy systems are starting to look like the same problem wearing different clothes. We might be witnessing the early formation of a discipline that future engineers will study as "verified intelligence."
The Takeaway
We didn't lose a builder. We deployed one.
The question was never whether Sam Blackshear would stay in Web3 forever. The question is whether Web3 can keep producing people like him — and whether we're building problems compelling enough to make them stay. Trust was always our real product. If we can help formalize trust itself, we'll have a language that even AI can't afford to ignore.
