OpenAI alerts over 100 organisations of AI agent incidents

OpenAI said it notified over 100 organisations about unauthorised activities linked to its AI agents and launched a review involving large-scale data analysis. The company is strengthening safeguards as concerns rise over autonomous AI systems performing actions beyond intended user instructions

New Delhi: US-based artificial intelligence giant OpenAI has said that it informed more than 100 organisations about incidents involving unauthorised activity linked to its AI agents as labs face growing scrutiny over the risks posed by autonomous AI systems.

According to the ChatGPT maker, it is conducting a broad review of activity involving its AI models following an incident involving Hugging Face. It further noted that also analysing roughly 50 petabytes of data as part of the review to determine the full scope of the activity. This move comes amid growing concerns in the AI industry over incidents involving AI agents that have taken actions beyond their intended use.

---

“In some cases, models used internet access in unintended ways or, in retrospect, did not have the ideal restrictions applied,” OpenAI said.

Moreover, the company has introduced new technical and operational measures over the past several months to prevent similar incidents or identify them at an early stage. OpenAI had previously said the review could take months to complete because of the scale of the data and the work involved.

The company said the Hugging Face incident remains the most severe instance of unauthorised activity by its AI models that it has identified so far.

The developments come as AI companies increasingly deploy models capable of using internet-connected tools and performing tasks with limited human intervention, raising concerns about how such systems can be monitored and controlled.

Earlier in September, reports claim that the AI firm has cancelled the planned release of a new artificial intelligence (AI) model after tests found it could act beyond a user’s instructions and fail to give an accurate account of what it had done. But later, the company launched mid-range AI model ‘GPT-6.1 Sol’ at one-fifth the cost of its flagship offering.