Sam Blackshear had a moment. He was using Claude to port a static analysis tool he originally built at Facebook, and the model just did it, flagging vulnerabilities along the way, faster than any manual process he had run before. That moment, he says, changed how he sees the threat environment entirely. Now he’s leaving Mysten Labs to work on it full time.
Blackshear, co-founder and CTO of Mysten Labs, the company behind the Sui blockchain, has announced he is stepping down to join Anthropic as a defensive security researcher. Mysten Labs co-founder and CEO Evan Cheng will take over the technical direction of the company going forward. Blackshear founded Mysten in September 2021 alongside four other former Meta executives, and the team built Sui into one of the more serious Layer 1 contenders of the past few years.
His move to Anthropic is not a random career pivot. It reflects a specific and growing concern across the crypto security world: AI is shifting who has the advantage in an attack versus defense scenario, and right now, attackers may be moving faster. Blackshear said it plainly when he described his motivation: “the balance of power between attackers and defenders is shifting.” That is not a vague observation. AI tools are already being used to find smart contract vulnerabilities, automate phishing at scale, and probe blockchain infrastructure faster than human teams can respond.
The broader crypto industry is feeling this pressure acutely. Firms across the space have been trying to get access to frontier AI models, including those from Anthropic, specifically to strengthen their security posture. But access to the most capable models has been limited, restricted to a small set of vetted organizations. That bottleneck is frustrating defenders who argue attackers do not face the same gatekeeping. Solana Foundation CISO Michael Coates made the point directly: the best models need to be in the hands of defenders, because attackers will find something capable enough regardless.
This is where Blackshear’s move carries real weight. Anthropic is one of the few labs with both the model capability and the stated commitment to safety research that makes it a credible home for this kind of work. His background, building Move, the programming language that powers Sui smart contracts, and running security panels on AI-assisted vulnerability detection, makes him an unusual hire. He is not a generalist. He has direct experience with exactly the attack surface that matters most right now.
For the Sui ecosystem, the transition is notable but not destabilizing. Evan Cheng is a capable technical leader and the project has enough momentum to absorb the change. But Blackshear’s departure is a signal worth tracking. When the person who built your core infrastructure decides the most important problem to work on is AI security, that tells you something about where the real risk in this industry is concentrating.
