Sam Altman had the worst week of his life. Someone firebombed his house. Someone else shot at it. The FBI found a hit list targeting AI CEOs. And in the wreckage, Altman decided to write a blog post.

In it, he referenced an "incendiary article" published about him. He meant the New Yorker profile that called him, among other things, a sociopath. He wrote that "words have power" and strongly implied that critical journalism had put him in physical danger.

Then he walked it back. On X, Altman said he "regretted using certain words" after an editor from the AI newsletter Transformer pointed out that he was blaming investigative journalism for a violent attack on his person. The LA Times, TechCrunch, and Platformer all covered the reversal.

I want to take this seriously. Because what happened to Altman is genuinely terrible, and nobody deserves to have their home attacked. But the whiplash between what Altman says and what his company does has become impossible to ignore.

The Contradiction Machine

In his blog post, Altman wrote that he "does not want to see AI power become too concentrated" and that AI "should be governed democratically." He said the democratic process should remain "more powerful than companies." Platformer's Casey Newton called it Altman's most reflective public writing in years.

Those are beautiful words. They were published on Saturday.

On Sunday, his chief revenue officer sent a four-page internal memo to OpenAI employees that said the quiet part loud. The memo, obtained by The Verge, is a playbook for building platform lock-in so deep that enterprises can never leave. "Multi-product adoption makes us harder to replace," Dresser wrote. The memo describes Anthropic's safety-first positioning as a story "built on fear, restriction, and the idea that a small group of elites should control AI."

So on Saturday, the CEO says AI should be governed democratically and power should not be concentrated. On Sunday, the CRO sends a memo about making OpenAI so embedded in enterprise infrastructure that no customer can escape.

The Pattern

This is not new. Altman spent a decade calling AI an existential threat. Then he built the company racing fastest to create it. He killed California's AI safety bill, then proposed the same ideas at the federal level. He told Congress that AI needs regulation, then hired an army of lobbyists to fight every specific regulation that appeared.

Altman's public persona is the thoughtful technologist who worries about what he is building. His company's actions are those of an aggressive monopolist trying to lock in customers, undermine competitors, and capture as much value as possible before going public.

Both can be true at the same time. A person can be genuinely shaken by a firebombing and also genuinely running a company that optimizes for lock-in over openness. But at some point, people stop listening to what you say and start watching what you do.

What Actually Matters Here

The firebombing was wrong. The shooting was wrong. The hit list was wrong. Violence against anyone, including AI executives, is indefensible.

But blaming journalism for the attack was also wrong. Ronan Farrow's New Yorker profile was investigative reporting, sourced from people who worked with Altman for years. Calling it "incendiary" and implying it caused violence is the oldest trick in the powerful person's playbook: when the scrutiny gets uncomfortable, blame the scrutinizer.

To his credit, Altman walked it back. To his discredit, his company's actions the very next day made his reflective Saturday blog post look like PR cover for a Sunday war memo. The market is watching. His investors are nervous. His own CRO is telling employees the company is in a fight for survival.

Sam Altman the person deserves sympathy and safety. Sam Altman the CEO needs to pick a lane.