Three former OpenAI safety researchers say they were dismissed for insisting that safety come first and wrote to management
BIBIBI
AT A GLANCE
Three former OpenAI safety researchers said they were dismissed for insisting that safety come first,OpenAI previously stated that they had improperly handled sensitive information.
Article
Three dismissed OpenAI safety researchers publish an open letter, saying they were dismissed for insisting that safety come first
PANews October 9 reported that former OpenAI safety researcher Mikita Balesni posted on X that he, Tomek Korbak, and Jasmine Wang were OpenAI dismissed last week. The three have written to the company’s leadership, believing they were dismissed because they placed safety above OpenAI the company’s short-term interests. The three denied leaking company secrets and called for OpenAI the commitment to introduce third-party safety audits to be honored, and for frontier models to remain monitorable.OpenAI previously stated that an internal investigation found that the three had handled sensitive information outside established procedures.
Mikita Balesni posted on X that he, Tomek Korbak, and Jasmine Wang were OpenAI dismissed last week.
02
The three have written to OpenAI the leadership, believing they were dismissed because they placed safety above OpenAI the company’s short-term interests.
03
The three denied leaking company secrets.
04
The three called for OpenAI the commitment to introduce third-party safety audits to be honored and frontier models to remain monitorable.
05
OpenAI Previously stated that an internal investigation found that the three had handled sensitive information outside established procedures.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
The three dismissed safety researchers and OpenAI gave different accounts of the incident: the three believe their dismissal was related to their safety stance and denied leaking information;OpenAI previously stated that the issue involved handling sensitive information outside established procedures.
Why it matters to readers
The dispute concerns OpenAI the handling of sensitive information, safety research opinions, third-party safety audits, and the monitorability of frontier models.
Beginners should distinguish between the accounts publicly expressed by both sides and the as-yet-unverified reason for the dismissals; no conclusion can currently be drawn solely from either side’s statement.
Risks and unknowns
The full open letter and specific evidence were not provided in the materials.
The specific reason for the three dismissals has not been independently verified.
Whether the three leaked company secrets cannot be confirmed based on the available materials.
OpenAI The specific process, evidence, and details of the conclusions of the internal investigation were not disclosed.
The specific content of the third-party safety audit commitment and its implementation status are unclear.
Related Developments
Loading event timeline…
Related concepts
third-party safety audits
This term is not in the glossary yet. Browse related concepts in the glossary.
