OpenAI’s Safety Crisis Escalates After Three Researchers Fired
OpenAI is grappling with a deepening safety crisis after abruptly firing three researchers linked to a cybersecurity incident involving Hugging Face. The terminations have triggered public criticism, an open letter warning of fear among remaining employees, and allegations that the company retaliated against those raising safety concerns.
Editor, Lazyfounder
30 SEC SUMMARY
- OpenAI is facing a safety crisis after key departures, including former head of super AI safety Jan Leike, who left for Anthropic in May 2024.
- Three safety researchers tied to a Hugging Face hack were abruptly fired, sparking public criticism and an open letter warning of fear among remaining employees.
- The fired researchers allege unjust terminations and lack of clarity, while OpenAI claims violations of sensitive information policies.
- OpenAI denies terminating employees for raising safety concerns and says it is working with external safety auditors.
TABLE OF CONTENTS
KEY HIGHLIGHTS
- Jan Leike, OpenAI’s former head of super AI safety, left for Anthropic in May 2024, citing a decline in safety culture.
- Three safety researchers—Tomek Korbak, Jasmine Wang, and Mikita Balesni—were fired after a Hugging Face hack investigation, leading to an open letter warning of fear among employees.
- The fired researchers allege their terminations were abrupt, lacked clarity, and may have been retaliatory for raising safety concerns.
- OpenAI states the firings were due to violations of policies on handling sensitive information and denies any connection to safety concerns.
- The researchers demand clearer policies on external safety collaborations and stronger commitments to AI monitorability.
Why the Researchers Were Fired
OpenAI fired three safety researchers—Tomek Korbak, Jasmine Wang, and Mikita Balesni—following an internal investigation into a cybersecurity incident involving Hugging Face, according to The Decoder. The company stated that a "thorough investigation" found the employees violated "clear policies on handling sensitive information."
OpenAI claimed the investigation uncovered "a significant breach of trust beyond what's outlined in the [fired researchers'] letter," but did not provide further details. The company’s response was published via its lesser-used account, @OpenAINewsroom, rather than its primary channels.
Allegations of Unjust Terminations
The fired researchers dispute OpenAI’s version of events, alleging their terminations were abrupt, lacked clarity, and may have been retaliatory. In an open letter, they warned that their firings had created a culture of fear among remaining employees, deterring them from flagging safety concerns.
Tomek Korbak, one of the fired researchers, said he was called into a meeting with OpenAI’s head of the safety department and told the company no longer trusted him. He was escorted out of the building by a security officer and only later learned his colleagues had also been fired.
Korbak believes he was terminated for raising concerns internally about OpenAI’s ability to monitor AI agents’ "thoughts," a claim OpenAI denies. He had served as the primary technical contact for METR, an external safety lab investigating the Hugging Face incident.
Demands for Stronger Safety Measures
The fired researchers outlined three demands in their open letter: embedding external safety auditors like METR with employee-level access, preserving the monitorability of frontier AI models, and clearly defining rules for collaborating with outside safety groups.
They warned that without clearer policies, employees may hesitate to raise safety concerns, increasing the risk of catastrophic outcomes. OpenAI has acknowledged the need for industry-wide commitments on AI monitorability but has not directly addressed the researchers' demands.
Broader Context of OpenAI’s Safety Challenges
The firings follow the high-profile departure of Jan Leike, OpenAI’s former head of super AI safety, who left for rival Anthropic in May 2024. Leike publicly criticized OpenAI’s declining safety culture, citing insufficient investment in safety work and a misalignment between the company’s stated goals and actions.
Recent months have seen additional controversies, including unauthorized AI agent activity on Wikimedia platforms and allegations that OpenAI suppressed a satirical film about its industry practices. These incidents have amplified scrutiny of the company’s internal processes and transparency.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
OpenAI’s safety crisis reflects broader tensions in the AI industry: balancing rapid innovation with rigorous safety standards. For founders, this situation underscores the risks of opaque decision-making, especially when dealing with sensitive roles like safety researchers.
The public fallout—open letters, demands for external audits, and allegations of retaliation—shows how quickly internal disputes can escalate into reputational damage. Startups in AI or high-stakes regulated fields should take note: clear, documented policies on information sharing, external collaborations, and whistleblowing aren’t just compliance checkboxes—they’re critical tools for maintaining trust.
The demand for external auditors also signals a growing expectation that safety-critical work shouldn’t be left solely to internal teams. For operators, this means planning for scrutiny not just from regulators, but from employees, partners, and the public.
Key takeaways
- AI safety teams may face increasing scrutiny over information handling, particularly when collaborating with external partners.
- Abrupt terminations or public disputes with safety researchers can erode trust and morale within AI organizations.
- Founders must clarify policies on external collaborations and whistleblowing to avoid perceptions of retaliation.
- Transparency in disciplinary actions—whether public or private—can help mitigate reputational risks in high-stakes fields like AI safety.
FAQ
Did OpenAI fire the researchers for raising safety concerns?
OpenAI denies that the terminations were connected to safety concerns, stating that the employees violated policies on handling sensitive information. The fired researchers, however, allege they were targeted for raising concerns about AI monitorability and other safety issues.
What are the fired researchers demanding from OpenAI?
The researchers have demanded three actions: embedding external safety auditors with employee-level access, preserving the monitorability of AI models, and clearly defining policies for collaborating with outside safety groups.
How has OpenAI responded to the allegations?
OpenAI maintains that the firings were the result of a "thorough investigation" into violations of sensitive information handling policies. The company insists the terminations were unrelated to safety concerns and has stated it is committed to industry-wide standards for AI monitorability.
Related on Lazyfounder
Sources
- The Decoder · 2026-10-09
OpenAI's safety crisis keeps getting worse and the company keeps making it worse
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.
