Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsJan Leike left OpenAI on May 17, 2024, after publicly describing a breakdown over priorities, then announced on May 28 that he had joined Anthropic. The move transferred a prominent alignment researcher from OpenAI’s Superalignment effort to an Anthropic team focused on scalable oversight, weak-to-strong generalization and automated alignment research.
When did Jan Leike leave OpenAI and join Anthropic?
| Date | Event |
|---|---|
| July 2023 | OpenAI launched its Superalignment effort, naming Ilya Sutskever and Jan Leike as co-leads. |
| May 17, 2024 | Leike’s last day at OpenAI, according to his statement and contemporaneous reporting by The Guardian. |
| May 28, 2024 | TechCrunch reported that Leike had joined Anthropic to lead a new research group. |
| September 27, 2026 | Leike’s personal biography identified him as lead of Anthropic’s Alignment Science team. That is a current self-description, not a guarantee that his role will remain unchanged. |
Why did Jan Leike leave OpenAI?
Leike attributed his departure to disagreements with OpenAI leadership about company priorities. In statements reported by The Guardian on May 18, 2024, he said he wanted more resources for safety, social impact, confidentiality and security in next-generation models.
He wrote that “safety culture and processes have taken a backseat to shiny products,” called building smarter-than-human machines “an inherently dangerous endeavour,” and said “OpenAI must become a safety-first AGI company.” Those are Leike’s criticisms and assessments. The available evidence does not independently establish why OpenAI made particular resource or product decisions, nor does it prove that safety work stopped.
What was OpenAI’s Superalignment Team?
OpenAI introduced Superalignment in July 2023 as a technical effort to address the problem of aligning superintelligent systems with human intent. Sutskever and Leike were announced as co-leads.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →#1 Best Overall
OpenAI said, “Our goal is to solve the core technical challenges of superintelligence alignment in four years.” It also said it planned to dedicate 20% of the compute it had secured to date to the work over the following four years. That figure and deadline were launch commitments, not independently verified spending or a demonstrated result.
The announcement distinguished Superalignment from safety work on current models and other AI risks. It described a research program, not a completed safety system or a comparative measure of company performance.
What will Jan Leike work on at Anthropic?
TechCrunch reported that Leike’s new Anthropic group would work on three areas:
Scalable oversight
Scalable oversight seeks ways for people, or systems supervised by people, to evaluate increasingly capable AI when direct human checking does not scale.
Recommended Free Tools
Rank #3
Weak-to-strong generalization
This research examines whether a less capable supervisor can reliably guide or improve a more capable model.
Automated alignment research
Automation applies AI systems to parts of the alignment research process, with the aim of making safety investigations more effective as model capabilities grow.
Leike’s biography also lists jailbreak robustness and describes the broader question as how to train AI systems to follow human intent on tasks that are difficult for humans to evaluate directly.
How do OpenAI and Anthropic’s documented safety approaches differ?
The public material supports a comparison of stated structures and research remits, not a ranking of which company is safer.
Free tools Windows power users keep installed
One-click scans. No signup required.
| Question | OpenAI snapshot | Anthropic snapshot |
|---|---|---|
| Research remit | Superalignment was announced as a four-year effort focused on the technical alignment of superintelligent systems, with a planned 20% allocation of secured compute. | Leike’s reported group covers scalable oversight, weak-to-strong generalization and automated alignment research; his biography also mentions jailbreak robustness. |
| Governance and assurance | On May 28, 2024, the Associated Press reported that OpenAI had announced a safety committee with a board advisory role and a planned review of processes and safeguards. | Anthropic’s May 20, 2024 Responsible Scaling Policy reflections discussed threat modeling, evaluations, safeguards and safety-assurance work across teams including Alignment Science and Frontier Red Team. |
| Evidence status | These announcements and statements do not provide an independently measured, like-for-like safety outcome or company ranking. | |
Anthropic’s policy reflections included pre-deployment testing in cybersecurity and chemical, biological, radiological and nuclear (CBRN) domains, as well as model-autonomy evaluations. They describe Anthropic’s broader safety program at that time; they are not proof of a specific result from Leike’s team.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What does the move mean for AI safety?
The move matters in two connected ways. It is a personnel change involving a former co-lead of OpenAI’s flagship superintelligence-alignment effort, and it continues a technical agenda centered on supervising systems whose behavior may be hard for humans to judge directly.
It does not, by itself, show that Anthropic has achieved a particular safety breakthrough, that OpenAI abandoned alignment, or that either company has a verified lower-risk model. Leike’s allegations should remain attributed to him, while company plans and policy documents should be read as stated intentions and program descriptions rather than outcome statistics.
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.
