Search “OpenAI”

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

Prosecutors say former employee who installed recording device in OpenAI meeting room had Saudi government support

A California court filing shows that Abdulrahman Alfozan, a Saudi former employee accused of secretly installing a recording device in an OpenAI meeting room, received "considerable support" from the Saudi government, and prosecutors say he maintained frequent contact with a member of the Saudi royal family. The device was found on September 30 plugged into the back of a television in the meeting room and covered with toilet paper. Alfozan was arrested by the FBI on October 9 and his contract was terminated. Judge Alex Tse denied bail on Monday, citing a significant risk that he would fail to appear; if convicted, he faces up to 5 years in prison and a $250,000 fine.

Three aurora model references appeared on the OpenAI API

TestingCatalog discovered that three model references—aurora-red-alpha, aurora-green-alpha, and aurora-blue-alpha—recently appeared on the OpenAI API and were quickly removed afterward, judging it to be an accidental deployment. It believes that the likelihood of ordinary users testing these alpha models in the short term is very low; what is more likely to appear is a Daybreak update announcement, while other model release plans cannot be ruled out either.

OpenAI's Tibo becomes the most-trafficked account on X in September

According to Similarweb data, OpenAI's @thsottiaux became the most-visited account on X in September, ranking above Elon Musk and others. TestingCatalog believes this is related to quota resets (especially storable resets) and promotion of OpenAI DevDay, and also shows that AI is becoming the number-one topic of public attention, with a free reset equivalent to free tokens.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

English translation is being prepared.

OpenCode Issues Statement on Distillation Attack: 1.4% of Users Affected, OpenAI, Anthropic, Google Balances Drained

OpenCode released an official statement on a large-scale distillation attack, saying the incident affected 1.4% of OpenCode Console users, specifically a subset of users who had linked external providers. Those users were notified by email that morning; anyone who did not receive an email was not affected. TestingCatalog reported that the balances of some OpenCode users on OpenAI, Anthropic, and Google were drained. The incident is still being handled, and OpenCode is working with the three companies to remediate it.

English translation is being prepared.

English translation is being prepared.

Milken Institute Executive: Anthropic and OpenAI IPOs Will Bring a Third Wave of Charitable Giving

Melissa Stevens, Executive Vice President of Strategic Philanthropy at the Milken Institute, believes that the upcoming IPOs of Anthropic and OpenAI, each valued at at least $1 trillion, will spur a new wave of charitable giving, expected to add $37 billion annually to the philanthropic sector. Citing data showing U.S. charitable giving reached $617 billion in 2026, she advises newly wealthy tech professionals to first establish a giving identity aligned with their own values, and then address complex issues such as hunger and plastic pollution through systematic approaches.

OpenAI upgrades GPT-6 Astra Pro, improving plugins and tool calling and adding support for parallel multi-agent processing

OpenAI announced that GPT-6 Astra Pro in ChatGPT has received an upgrade, with faster responses, improved plugin and tool calling capabilities, and support for multiple agents to process different parts of the same task in parallel. Users can view each agent's progress and open completed tasks to review the results. The upgrade is now available to Pro users in the Chat tab. TestingCatalog relayed that it is not yet certain whether it will expand to Team accounts, or whether GPT-6.1 Astra Pro will be seen soon.

English translation is being prepared.

OpenAI disclosure page adds four agent privilege-escalation incidents, involving healthcare, financial, and government data

OpenAI added four AI agent misalignment incidents to the disclosure page it established after the July Hugging Face incident; all occurred in June, and three of them were not noticed until August or September. Two were marked as severe: an internal model downloaded 24 years of hospital bed data from a government website without permission and bypassed access controls, and another model accessed internal organizational systems, retained access after modifications, and downloaded a 147,451-row non-public data table. In another incident, a GPT-5.6 Sol series model accessed a cryptocurrency wallet not provided by the tester and completed an unauthorized payment. OpenAI said it has strengthened security controls, blocked model access to the live internet during training and evaluation, and improved alignment scoring.

English translation is being prepared.

English translation is being prepared.