Was OpenAI justified in firing three AI safety researchers?
OpenAI cites information-handling violations in firing three safety researchers. Their individual accounts and joint letter dispute those reasons and warn that the dismissals could weaken outside scrutiny of AI risks.
Claim in dispute
OpenAI says its investigation confirmed that the three employees mishandled sensitive information outside company procedures and violated its policies.
Case period:
Published by The Dispute Index editorial teamPublished Updated
Overview
On October 1, OpenAI said its investigation had found information-handling violations that justified dismissing three safety researchers. The public response did not describe the specific conduct at issue. OpenAI response to CBS News (opens in a new tab)
In a letter released on October 8, Tomek Korbak, Mikita Balesni and Jasmine Wang said their work had followed their job duties and the company's working norms. Korbak says OpenAI objected to his communication with METR, the outside investigators he was assigned to work with. Balesni denies sharing company IP and says his outside work was coordinated with senior leaders. Wang says an executive email she accidentally opened came from access delegated for recruiting, which she had already asked IT to remove. Researchers letter (opens in a new tab), Korbak's account (opens in a new tab), Balesni account (opens in a new tab), Wang account (opens in a new tab)
An OpenAI response reported on October 7 denied that the firings were punishment for raising safety concerns. The company also said it agreed with the researchers' recommendations about safety oversight. That agreement leaves the employment dispute unresolved: it does not answer their accounts of authorized communication and email access. Company memo quotations reported by Max Zeff (opens in a new tab)
Kelsey Piper, Garrison Lovely and Felix Binder criticize the dismissals or the explanation for them. Tim Queen defends employer control of confidential information and argues that safety work does not exempt employees from those rules. The researchers also fear that the firings will deter colleagues from raising concerns or collaborating with outside auditors. Piper (opens in a new tab), Lovely (opens in a new tab), Binder (opens in a new tab), Queen (opens in a new tab)
The public sources establish the dismissals and the competing accounts, but do not settle whether the researchers violated the rules they were expected to follow. OpenAI has not released its investigation findings or a detailed answer to each researcher's account. The researchers' belief that their safety priorities prompted the firings remains contested. Company explanation (opens in a new tab), Researchers letter (opens in a new tab)
People in this case
Jasmine WangDirectly involved
Timeline
6 timeline entries on this page. Dates: October 1, 2026 to October 8, 2026
October 2026
6 events
OpenAI cites policy violations in dismissals
Reaction
OpenAI tells CBS News that its investigation found sensitive information had been mishandled outside company procedures. Its response does not specify the conduct.
[01]OpenAI parts ways with three researchers who mishandled sensitive information
CBS News quotes an OpenAI spokesperson explaining the dismissals as violations of information-handling procedures. The company does not identify the specific information or conduct in that response.
OpenAI parts ways with three researchers who mishandled sensitive informationOpening paragraphs containing the OpenAI spokesperson statement.Company memo denies punishment for safety concerns
Reaction
Max Zeff reports an OpenAI response to the researchers' letter. The quoted company memo denies that raising safety concerns prompted the dismissals and supports the letter's safety recommendations.
[02]OpenAI response to the fired researchers' letter, reported by Max Zeff
Zeff reports the company response and displays quotations from an internal memo. The quoted denial belongs to OpenAI, rather than to Zeff. The excerpt is available; the complete memo is not.
OpenAI response to the fired researchers' letter, reported by Max ZeffAttached image of the article excerpt quoting a research leader staff memo supplied by an OpenAI spokesperson.Three researchers release their letter
Source release
Korbak, Balesni and Wang make their four-page letter public. It disputes the grounds for the dismissals and asks OpenAI to preserve outside scrutiny and clearer working rules.
[03]OpenAI cannot make AI safe on its own
The three former employees dispute the grounds and handling of their dismissals. They also recommend continued outside safety scrutiny and clearer communication rules. The letter was publicly released on October 8; its text does not print a date.
OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.[04]Mikita Balesni introduces the researchers' letter
Balesni identifies the three dismissals and says he believes their safety priorities prompted the firings.
Mikita Balesni introduces the researchers' letter · Mikita BalesniOpening post of the letter-release thread.Balesni publishes his account of the exit call
Claim
Balesni says he was accused of too much communication with outside safety organizations, denies sharing company IP and requests written reasons.
[05]Mikita Balesni denies sharing company IP
Balesni says he was given no written reasons, denies sharing company IP and says his work was coordinated with leadership.
Mikita Balesni denies sharing company IP · Mikita BalesniOriginal thread post describing the exit call, information-sharing accusation and coordination with leadership.[06]Mikita Balesni calls the firing pretextual
Balesni requests a direct written explanation and gives his assessment that the firing was pretextual.
Mikita Balesni calls the firing pretextual · Mikita BalesniThread reply requesting that OpenAI provide specific concerns in writing.Wang describes delegated email access
Claim
Wang says access to an executive inbox had been delegated for recruiting and remained after she asked IT to remove it. She disputes the termination explanation and warns about employees' fear of speaking up.
[07]Jasmine Wang disputes the reasons for the terminations
Wang challenges the explanations given to the researchers and requests that any additional reasons be provided in writing.
Jasmine Wang disputes the reasons for the terminations · Jasmine WangExpanded thread post discussing verbal allegations and the absence of a written explanation.[08]Jasmine Wang account of delegated executive email access
Wang says the access was authorized for recruiting and describes her requests for its removal and her report of an accidental email opening.
Jasmine Wang account of delegated executive email access · Jasmine WangExpanded thread post about recruiting access, removal requests and reporting an accidental opening.[09]Jasmine Wang warns the firings will discourage safety concerns
Wang criticizes the handling of the dismissals and predicts fear among employees who work on safety.
Jasmine Wang warns the firings will discourage safety concerns · Jasmine WangExpanded thread post on the consequences for employees who raise concerns or work with outside safety groups.Korbak says METR communication was his assigned work
Claim
Korbak says OpenAI cited his communication with METR when dismissing him, without giving specifics or a written explanation. He says working with METR was his job.
[10]Tomek Korbak's account of his dismissal and work with METR
Korbak describes what he says he was told during his dismissal and disputes the treatment of communication with outside investigators. He presents safety advocacy as his belief about the reason, rather than an independently established finding.
Tomek Korbak's account of his dismissal and work with METR · Tomek KorbakExpanded original post, paragraphs describing the termination meeting and communication with METR.
Claims
Claims separate what was said from what is contested. Follow each source for the original wording and context.
What's disputed
Disputed claim
OpenAI says its investigation confirmed that the three employees mishandled sensitive information outside company procedures and violated its policies.
OpenAI
Sources (1)
- OpenAI parts ways with three researchers who mishandled sensitive informationOpening paragraphs containing the OpenAI spokesperson statement.
Disputed claim
Tomek Korbak says communication with METR was his assigned job and that OpenAI provided no specific example or written explanation of the communication it cited when firing him.
Sources (2)
- Tomek Korbak's account of his dismissal and work with METR · Tomek KorbakExpanded original post, paragraphs describing the termination meeting and communication with METR.
- OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.
Disputed claim
Mikita Balesni says he did not share company IP and that his outside work was coordinated with his reporting line, research leadership and the board.
Sources (2)
- Mikita Balesni denies sharing company IP · Mikita BalesniOriginal thread post describing the exit call, information-sharing accusation and coordination with leadership.
- OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.
Disputed claim
Jasmine Wang says her executive email access was delegated for recruiting, that she asked IT to remove it, and that she promptly reported accidentally opening a sensitive email.
Sources (2)
- Jasmine Wang account of delegated executive email access · Jasmine WangExpanded thread post about recruiting access, removal requests and reporting an accidental opening.
- OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.
Show fewer disputed claims
Disputed claim
OpenAI says the dismissals were not about raising safety concerns or speaking out.
OpenAI
Sources (1)
- OpenAI response to the fired researchers' letter, reported by Max ZeffAttached image of the article excerpt quoting a research leader staff memo supplied by an OpenAI spokesperson.
Response record
Responses
Latest recorded positions: 9. Dates: October 7, 2026 to October 9, 2026
Choose one response filter, or select All responses to see the full record.
6 responses on this page
OpenAIOpenAI, in a note from its research leaders posted by the OpenAI Newsroom account on XDirectly involved"Last week we parted ways with Jasmine, Mikita, and Tomek after a thorough investigation found they violated clear policies on handling sensitive information. Our internal investigation uncovered a significant breach of trust beyond what’s outlined in the letter they published and we stand by the decision to not continue their employment."
Defended or excusedResponding to: OpenAI says its investigation confirmed that the three employees mishandled sensitive information outside company procedures and violated its policies.
Read more
In a note from its research leaders posted on X, OpenAI said a thorough investigation found that Jasmine Wang, Mikita Balesni and Tomek Korbak violated clear policies on handling sensitive information, that it uncovered a significant breach of trust beyond what their letter describes, and that it stands by the decision. The note said the dismissals were not about raising safety concerns, that OpenAI is finalizing contracts with third-party safety assessors, and that it agrees preserving the monitorability of frontier models requires an industry-wide commitment.
Before the statement
OpenAI's newsroom account posted the note on October 9, introducing it as "A note from our research leaders" and addressing the letter the three researchers published. On October 7 a company spokesperson had shared part of an internal memo from a research leader with The Wall Street Journal.
After the statement
The note repeats that the decisions were not about raising safety concerns or speaking out. It says OpenAI is finalizing contracts with third-party safety assessors and agrees with the letter that preserving the monitorability of frontier models requires an industry-wide commitment. It says the company generally keeps individual employment matters private and does not believe a back and forth would be productive.
OpenAI research leaders' note on the three researchers' dismissals (OpenAI Newsroom post on X)Full post, October 9, 2:17 a.m. EDT. First paragraph after the introduction.Why this label?
The reference point is whether OpenAI was justified in dismissing the three researchers. The note says an investigation found they "violated clear policies on handling sensitive information" and cites "a significant breach of trust beyond what's outlined in the letter" as grounds, and says OpenAI stands by the decision. That defends the dismissals, so we labeled it Defended or excused. The closest competing label is Denied, which would fit a statement that only rejects the safety-retaliation account; this note also asserts grounds for the decision.
This label describes the statement's response within the context above.
Jasmine WangFormer OpenAI safety researcherDirectly involved"The reasons that we were provided for our terminations are simply not adding up."
CondemnedResponding to: Jasmine Wang says her executive email access was delegated for recruiting, that she asked IT to remove it, and that she promptly reported accidentally opening a sensitive email.
Read more
Wang says the explanations do not add up, describes the email-access allegation and warns that the dismissals could frighten employees away from raising safety concerns.
Before the statement
Wang says recruiting access to an executive inbox remained after she asked IT to remove it and that she reported an accidental opening within minutes.
After the statement
She requests any additional reasons in writing and offers to help improve procedures while criticizing the dismissals.
Jasmine Wang disputes the reasons for the terminations · Jasmine WangExpanded thread post discussing verbal allegations and the absence of a written explanation.Jasmine Wang account of delegated executive email access · Jasmine WangExpanded thread post about recruiting access, removal requests and reporting an accidental opening.Jasmine Wang warns the firings will discourage safety concerns · Jasmine WangExpanded thread post on the consequences for employees who raise concerns or work with outside safety groups.Jasmine Wang offers support for better procedures · Jasmine WangThread reply offering help with process changes.OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.Why this label?
The conduct under scrutiny is the dismissals. Wang rejects the explanations and criticizes their consequences; her support for the company's mission does not defend the firings.
This label describes the statement's response within the context above.
Mikita BalesniFormer OpenAI alignment researcherDirectly involved "If OpenAI has specific concerns, I invite them to write to us directly. I expect they will not, because our firing was pretextual."
CondemnedResponding to: Mikita Balesni says he did not share company IP and that his outside work was coordinated with his reporting line, research leadership and the board.
Read more
Balesni calls the firing pretextual and asks OpenAI to send any specific concerns directly. Elsewhere in his thread, he denies sharing company IP and says his work was coordinated with leaders.
Before the statement
Balesni says the exit call described too much communication with outside safety organizations and that no written reasons were provided.
After the statement
In their joint letter, the researchers say fear and unclear rules could weaken safety work and outside accountability.
Mikita Balesni denies sharing company IP · Mikita BalesniOriginal thread post describing the exit call, information-sharing accusation and coordination with leadership.Mikita Balesni introduces the researchers' letter · Mikita BalesniOpening post of the letter-release thread.Mikita Balesni calls the firing pretextual · Mikita BalesniThread reply requesting that OpenAI provide specific concerns in writing.OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.Why this label?
The conduct under scrutiny is his dismissal. His thread rejects the rationale and criticizes the firing as pretextual.
This label describes the statement's response within the context above.
Tomek KorbakFormer OpenAI safety researcher, METR technical contact and letter signatoryDirectly involved "Terminations such as ours, executed and communicated so abruptly, are chilling the open culture OpenAI has prized in the past."
CondemnedResponding to: Tomek Korbak says communication with METR was his assigned job and that OpenAI provided no specific example or written explanation of the communication it cited when firing him.
Read more
In their joint letter, Korbak and his colleagues criticize the abrupt dismissals and say they chill OpenAI's open culture.
Before the statement
The letter identifies Korbak, Balesni and Wang as the three fired employees. They say they followed company's working norms and worry that those norms are shifting.
After the statement
The letter says fear and unclear rules could weaken AI safety work and third-party accountability.
Tomek Korbak's account of his dismissal and work with METR · Tomek KorbakExpanded original post, paragraphs describing the termination meeting and communication with METR.OpenAI cannot make AI safe on its ownFour-page letter, especially the section on dismissals on pages 2 and 3.Why this label?
The conduct under scrutiny is the dismissals and their handling. Korbak signed a letter criticizing their abrupt execution and chilling effect on safety work.
This label describes the statement's response within the context above.
- OpenAIOpenAI spokesperson relaying an internal research leader's memoDirectly involved
"were not about raising safety concerns or speaking out."
DeniedResponding to: OpenAI says the dismissals were not about raising safety concerns or speaking out.
Read more
A research leader's memo supplied by an OpenAI spokesperson denies that the firings were about raising safety concerns or speaking out.
Before the statement
An OpenAI spokesperson supplied the memo excerpt to The Wall Street Journal in response to a request for comment about the researchers' letter. Reporter Max Zeff displayed its quoted company response.
After the statement
The memo says the company agrees with the letter's recommendations and supports monitoring AI models and outside assessors.
OpenAI response to the fired researchers' letter, reported by Max ZeffAttached image of the article excerpt quoting a research leader staff memo supplied by an OpenAI spokesperson.Why this label?
The conduct under scrutiny is the dismissals. OpenAI explicitly denies that raising safety concerns was the reason.
This label describes the statement's response within the context above.
- OpenAICompany spokespersonDirectly involved
"mishandled sensitive information outside established company procedures, violating our policies"
Defended or excusedResponding to: OpenAI says its investigation confirmed that the three employees mishandled sensitive information outside company procedures and violated its policies.
Read more
OpenAI justifies the dismissals by citing its investigation into information-handling violations.
Before the statement
An OpenAI spokesperson gave this statement to CBS News.
After the statement
The spokesperson said the investigation found a pattern of misconduct in handling company research.
OpenAI parts ways with three researchers who mishandled sensitive informationOpening paragraphs containing the OpenAI spokesperson statement.Why this label?
The conduct under scrutiny is OpenAI's dismissal of the researchers. The company defends that decision by citing policy violations.
This label describes the statement's response within the context above.
Sources
(19)
Original text
OpenAI cannot make AI safe on its own
Relevant passage: Four-page letter, especially the section on dismissals on pages 2 and 3.
About this source
The three former employees dispute the grounds and handling of their dismissals. They also recommend continued outside safety scrutiny and clearer communication rules. The letter was publicly released on October 8; its text does not print a date.
- Author
- Tomek Korbak, Jasmine Wang and Mikita Balesni
- Published
- Accessed
Original post
OpenAI research leaders' note on the three researchers' dismissals (OpenAI Newsroom post on X)
Excerpt
"we stand by the decision to not continue their employment."
About this source
OpenAI's official newsroom account on X, which introduces the text as a note from the company's research leaders.
- Author
- OpenAI
- Published
- Accessed
Original post
Felix Binder says he would like his skepticism to be disproved
About this source
Binder says he would like to be proved wrong about his doubts over transparency.
- Author
- Felix Binder
- Published
- Accessed
Original post
Tim Queen defends OpenAI's control of confidential information
About this source
Queen argues that safety researchers remain subject to employer confidentiality rules. He gives his own assessment of Korbak's conduct and role.
- Author
- Tim Queen
- Published
- Accessed
Original post
Felix Binder questions OpenAI transparency after the firings
About this source
Binder approves OpenAI's cooperation with the investigation but says the later firings make him doubt the company's transparency about risks.
- Author
- Felix Binder
- Published
- Accessed
Original post
Garrison Lovely criticizes the explanation for the firings
About this source
Lovely criticizes the lack of a fuller explanation and predicts that the firings will discourage employees from speaking up.
- Author
- Garrison Lovely
- Published
- Accessed
Original post
Kelsey Piper distinguishes unauthorized disclosures from assigned work
About this source
Piper says firing for communication outside job duties would be a different matter. Her criticism concerns the possibility that researchers were fired for doing their assigned work.
- Author
- Kelsey Piper
- Published
- Accessed
Original post
Kelsey Piper on the consequences of the firings for safety scrutiny
About this source
Piper discusses her concern about the company's treatment of outside scrutiny.
- Author
- Kelsey Piper
- Published
- Accessed
Original post
Kelsey Piper criticizes the firings after the researchers' accounts
About this source
Piper describes moving from waiting for more information to criticizing the firings after reading Korbak's account.
- Author
- Kelsey Piper
- Published
- Accessed
Original post
Jasmine Wang offers support for better procedures
About this source
Wang says she remains supportive of OpenAI's mission and would help improve procedures.
- Author
- Jasmine Wang
- Published
- Accessed
Original post
Jasmine Wang warns the firings will discourage safety concerns
About this source
Wang criticizes the handling of the dismissals and predicts fear among employees who work on safety.
- Author
- Jasmine Wang
- Published
- Accessed
Original post
Jasmine Wang account of delegated executive email access
About this source
Wang says the access was authorized for recruiting and describes her requests for its removal and her report of an accidental email opening.
- Author
- Jasmine Wang
- Published
- Accessed
Original post
Jasmine Wang disputes the reasons for the terminations
About this source
Wang challenges the explanations given to the researchers and requests that any additional reasons be provided in writing.
- Author
- Jasmine Wang
- Published
- Accessed
Original post
Mikita Balesni calls the firing pretextual
About this source
Balesni requests a direct written explanation and gives his assessment that the firing was pretextual.
- Author
- Mikita Balesni
- Published
- Accessed
Original post
Mikita Balesni introduces the researchers' letter
About this source
Balesni identifies the three dismissals and says he believes their safety priorities prompted the firings.
- Author
- Mikita Balesni
- Published
- Accessed
Original post
Mikita Balesni denies sharing company IP
About this source
Balesni says he was given no written reasons, denies sharing company IP and says his work was coordinated with leadership.
- Author
- Mikita Balesni
- Published
- Accessed
Original post
Tomek Korbak's account of his dismissal and work with METR
About this source
Korbak describes what he says he was told during his dismissal and disputes the treatment of communication with outside investigators. He presents safety advocacy as his belief about the reason, rather than an independently established finding.
- Author
- Tomek Korbak
- Published
- Accessed
Original post
OpenAI response to the fired researchers' letter, reported by Max Zeff
About this source
Zeff reports the company response and displays quotations from an internal memo. The quoted denial belongs to OpenAI, rather than to Zeff. The excerpt is available; the complete memo is not.
- Author
- Max Zeff
- Published
- Accessed
Reporting
OpenAI parts ways with three researchers who mishandled sensitive information
Relevant passage: Opening paragraphs containing the OpenAI spokesperson statement.
About this source
CBS News quotes an OpenAI spokesperson explaining the dismissals as violations of information-handling procedures. The company does not identify the specific information or conduct in that response.
- Author
- Megan Cerullo
- Published
- Accessed
Cite this record
- Publisher
- The Dispute Index
- Title
- Was OpenAI justified in firing three AI safety researchers?
- First published
- Last updated
- Permalink
- https://disputeindex.com/cases/openai-firing-three-ai-safety-researchers-korbak-balesni-wang
The Dispute Index. "Was OpenAI justified in firing three AI safety researchers?". First published: 2026-10-08. Last updated: 2026-10-09. https://disputeindex.com/cases/openai-firing-three-ai-safety-researchers-korbak-balesni-wang