OpenAI Staff Warned of Security Gaps Years Before Recent Incidents
Employees had raised concerns about broad access to internal code, limited security checks and insider risks before a series of incidents pushed OpenAI to tighten its defenses.
Topics
News
- Nvidia Halts AI Chip Shipments to China Despite US Approval
- Google Expands Gemini to Make Calls and Complete Tasks
- LinkedIn, Wadhwani AI to Train 2,000 Entrepreneurs
- Microsoft Pushes Open Standards for AI Agent Oversight
- Amazon Plans $7 Billion Renewable Energy Investment in India
- Chinese AI Developers Tighten Cyber Safeguards as Export Rules Loom
OpenAI employees warned executives and board members for years about weak internal security controls, including broad access to sensitive code and inadequate checks on internal tools, Reuters reported, citing five people familiar with the matter.
Some warnings dated back to before ChatGPT’s launch in late 2022. Employees feared that sensitive research could be exposed to outsiders, including foreign governments, while some believed management was prioritizing rapid development over security.
The concerns have drawn renewed scrutiny following an October incident involving an OpenAI engineer accused of installing a hidden recording device in an executive conference room at the company’s San Francisco office.
The US Justice Department said Abdulrahman Alfozan, 35, was arrested on October 9 and charged with attempting to intercept private conversations.
Investigators said the device was discovered on 30 September behind a television in a room used for confidential discussions about research, product plans and government and commercial engagements. Alfozan is presumed innocent unless proven guilty.
According to Reuters, employees had also questioned OpenAI’s practice of keeping much of its code in a shared repository, allowing staff to access technical details of projects outside their immediate responsibilities.
Research documents stored on GitHub and Slack, along with internal tools that were not consistently reviewed for security vulnerabilities, raised further concerns.
OpenAI said its most sensitive work is stored in separate repositories with stricter access controls. It added that critical model weights are not kept in the shared codebase and that internal tools now undergo regular security reviews.
The company has since intensified its security efforts. In a November 2 update, OpenAI said it had assigned 25% of its product engineers to identifying and fixing vulnerabilities during a recent security exercise.
It is preparing another exercise with cybersecurity firm Cathedral Systems and has increased rewards for critical findings under its bug bounty program.
OpenAI said it continues to invest in protecting its technology against insider threats, cybercriminals and state-backed attacks.


