Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
OpenAI hired Dylan Scandinaro from Anthropic as its Head of Preparedness, Sam Altman announced on February 3, 2026. The role is meant to help the company prepare for and mitigate serious risks from increasingly capable AI models. The appointment puts a prominent safety-related hire inside OpenAI; it does not, by itself, show that Scandinaro can block a model release or that the company’s safety governance has changed.
Who is Dylan Scandinaro?
Scandinaro had worked at Anthropic in a role described in contemporaneous coverage as related to AGI safety. That description is reported rather than a confirmed formal Anthropic job title. Coverage of the appointment also traces his earlier career to Unity Technologies, Everyday Robots and DeepMind, and describes product-management experience before Anthropic; those details were attributed to his LinkedIn profile rather than an official biography. The Verge’s account of the hire and GIGAZINE’s report provide the available career context.
The public information establishes his new title and broad remit, but does not provide a detailed biography of his research, published evaluations or policy work. It is therefore more accurate to describe him as a safety and preparedness leader than to assign him a specific research specialty that the available reporting does not establish.
What does “Head of Preparedness” mean?
Preparedness is narrower than the whole field of AI safety and different from routine content moderation. In this context, it concerns whether a model could create unusually severe risks as its capabilities grow, how those risks should be tested, and what safeguards or response plans should be in place. Depending on a company’s framework, that work can involve evaluating dangerous capabilities, coordinating red-team or external testing, setting escalation procedures, and advising leadership about deployment conditions.
#1 Best Overall
- OpenAI develops a unified AI platform designed to act as a proactive assistant for daily life. OpenAI provides advanced reasoning, multimodal search, and agentic workflows that simplify complex tasks to expand what individuals and teams can build.
- OpenAI is for entrepreneurs, teachers and students, artists and scientific researchers. OpenAI is for those building AI-native projects, creating art, or working with data workflows who need a digital agent to accelerate learning and project completion.
- Lightweight, Classic fit, Double-needle sleeve and bottom hem
Those are the kinds of responsibilities associated with a preparedness function; public reporting on Scandinaro’s appointment does not set out his specific job description, team size, budget or authority. OpenAI has not, in the cited appointment coverage, established that he can independently delay or veto a release. Preparedness also should not be taken to mean that one executive handles every safety issue, including privacy, bias, emotional dependency or everyday misuse.
What OpenAI and Scandinaro said about the appointment
Altman said Scandinaro would lead OpenAI’s efforts to prepare for and mitigate serious risks. Scandinaro’s own public statement framed the work around the growing potential benefits of AI as well as extreme risks, and the limited time available to prepare. These statements describe the stated mission, not the decision-making structure behind it. GIGAZINE’s report summarizes the announcement, while The Verge covers the appointment.
Rank #2
- For men, women, Python programmers, computer science teachers, video players, math teachers, nerds, geeks, sci-fi fantasy fans. Ideal for back to school, Christmas, Halloween, graduation, meetings, Pi day and computer conferences.
- Ideal for nerds, computer geeks, students, programmers, hackers, sci-fi fans, retro video game console fans. Nerdy OpenAI Artificial Intelligence Computer Programmer
- Lightweight, Classic fit, Double-needle sleeve and bottom hem
Why hiring from Anthropic matters
Anthropic has cultivated a public profile around AI safety, including constitutional AI and model-risk evaluation. Recruiting an employee from a rival with that reputation may give OpenAI access to experience developed in a different frontier-lab environment. The available reporting does not establish that Scandinaro was hired to reproduce a particular Anthropic method, nor does it establish that he left because of a disagreement or dissatisfaction.
The move is part of a broader, two-way flow of talent. In 2024, former OpenAI safety leader Jan Leike joined Anthropic to lead a new team focused on superalignment, as reported by TechCrunch and Axios. In May 2026, TechCrunch reported that OpenAI co-founder Andrej Karpathy joined Anthropic’s pre-training team. That move involved a different technical area, but it illustrates how personnel circulate between the labs. Safety, preparedness and evaluation expertise have become part of that competition alongside research and engineering talent.
Rank #3
- For Men, Women, Python programmers, computer science teachers, video game players, math teachers, nerds, geeks, sci-fi fantasy fans. Great for back to school, Christmas, Halloween, graduation, conventions, pi day, and computer conferences.
- Great for nerds, computer geeks, students, programmers, hackers, sci-fi fans, retro video game consoles fans. Nerdy OpenAI Artificial Intelligence Computer Programmer
- Lightweight, Classic fit, Double-needle sleeve and bottom hem
Anthropic was also continuing to hire for safety-related work in 2026: Axios reported roles spanning cybercrime, chemical and explosive misuse, nuclear-weapons-related risks and other high-severity threats. That reporting is a reminder that Scandinaro’s departure does not establish that Anthropic’s safety effort ended. Separately, TIME reported that Anthropic dropped an earlier safety pledge concerning model releases and risk thresholds. The report offers context about pressures facing safety commitments across the industry, but it does not connect that change to Scandinaro’s move.
What is not yet clear about his authority
The key governance question is not only what preparedness work Scandinaro leads, but how its findings affect decisions. Appointment coverage does not establish his reporting line, whether his recommendations are binding, or who makes the final call when testing identifies a serious risk. It also does not specify whether the role sits outside the product chain or has board-level oversight.
Rank #4
- For Men, Women, Python programmers, computer science teachers, video game players, math teachers, nerds, geeks, sci-fi fantasy fans. Great for back to school, Christmas, Halloween, graduation, conventions, pi day, and computer conferences.
- Great for nerds, computer geeks, students, programmers, hackers, sci-fi fans, retro video game consoles fans. Nerdy OpenAI Artificial Intelligence Computer Programmer
- Lightweight, Classic fit, Double-needle sleeve and bottom hem
To assess whether the position changes how OpenAI deploys models, readers would need public evidence on:
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall- Authority: whether the preparedness lead can delay a release or only advise decision-makers.
- Resources and scope: the size of the team, its budget, and which severe-risk categories it evaluates.
- Standards: whether risk thresholds are published, measurable and independently auditable.
- Escalation and follow-through: who acts when a model crosses a threshold, and whether the company reports cases in which assessments changed a deployment decision.
- Continuity: whether the function and its authority remain in place through later organizational changes.
A summer 2026 assessment by the Future of Life Institute criticized both OpenAI and Anthropic for weak or unclear decision authority around safety frameworks, while noting OpenAI’s comparatively broad risk-assessment and external-testing activity. That is an outside assessment, not documentation of Scandinaro’s powers. The institute’s Summer 2026 AI Safety Index provides that broader evaluation.
Best Value
- OpenAI develops a unified AI platform designed to act as a proactive assistant for daily life. OpenAI provides advanced reasoning, multimodal search, and agentic workflows that simplify complex tasks to expand what individuals and teams can build.
- OpenAI is for entrepreneurs, teachers and students, artists and scientific researchers. OpenAI is for those building AI-native projects, creating art, or working with data workflows who need a digital agent to accelerate learning and project completion.
- Lightweight, Classic fit, Double-needle sleeve and bottom hem
What the reported $555,000 figure does—and does not—tell you
Bloomberg reported that the OpenAI safety job was listed at $555,000. The figure is reported compensation information tied to a job listing, not confirmation of Scandinaro’s final pay. The available reporting does not establish whether it was base salary or another compensation measure, what location or pay range it covered, or whether equity, bonuses and benefits were included. Bloomberg’s report is the source of the listing figure; it is not evidence that OpenAI outbid Anthropic or paid Scandinaro that amount.
What later developments show—and what they do not
In August 2026, Axios reported that OpenAI slowed the release of an upcoming model, Astra, after concerns about its potential “critical” cyber capabilities. The report is an example of why preparedness processes can matter in practice: capability assessments may affect when a model is released. It does not establish that Scandinaro personally initiated or directed that decision.
The appointment is meaningful as evidence that OpenAI recruited an employee from a rival lab for a senior preparedness role, at a time when frontier companies were competing for safety expertise. Whether it amounts to a safety upgrade depends on the institutional details that remain unestablished in the public appointment coverage: who can act on the role’s findings, under what standards, and with what accountability.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
