OpenAI is hiring a new head of preparedness

If you're looking for a gig that pays $555,000 base plus equity, OpenAI is hiring. Just be prepared to dive into the deep end on day one.

OpenAI CEO Sam Altman took to X to announce the search for a new Head of Preparedness. The executive will be tasked with running the company's Preparedness Framework, tracking frontier capabilities, and mitigating severe harms before models spiral out of control.

Real exploits and mental health fallout

Altman didn't mince words about why the role is suddenly so urgent. Frontier models are no longer just hallucinating trivia; they are getting good enough at computer security that they are actively finding critical software vulnerabilities.

Then there is the human fallout. Altman noted that 2025 offered a preview of models impacting mental health. The company is currently facing lawsuits alleging ChatGPT reinforced delusions, exacerbated isolation, and contributed to suicides, though OpenAI says it is actively training systems to recognize emotional distress.

The job description also explicitly highlights risks around releasing biological capabilities and ensuring safety around systems that can self-improve.

A team with high turnover and shifting rules

OpenAI first set up its Preparedness team back in 2023 to evaluate catastrophic risks ranging from phishing to nuclear threats. But the track record since then has been rocky.

Less than a year into the job, inaugural head Aleksander Madry was reassigned to a reasoning-focused role. Several other top safety executives have either walked out the door or pivoted to non-safety work.

Making things trickier, OpenAI recently updated its Preparedness Framework to include an interesting loophole: the company noted it might adjust its safety requirements if a rival lab pushes out a high-risk model without similar safeguards.

Why it matters

This isn't about sci-fi extinction threats anymore; it's about product liability and weaponized code. Between wrongful death lawsuits and models uncovering zero-day exploits, OpenAI is feeling acute pressure from both regulators and the courts.

We think the $555k package reflects just how thankless this job might actually be. The new hire will have to balance real defensive security against aggressive commercial roadmaps. And with OpenAI's own rulebook admitting it may loosen safety gates if competitors cut corners, anyone taking this desk faces an uphill battle to prove they hold real veto power rather than just corporate cover.