OpenAI’s New Security Report Reveals Latest AI Threats

OpenAI launched its latest threat report, ​​​​“Disrupting Malicious Uses of AI,” on Tuesday, revealing how hackers have been using AI for cyberattacks. Malicious actors have been using ChatGPT to assist in their operations, applying different strategies.

According to OpenAI’s report, the recent threat analyses, which the startup began issuing in February, have helped it understand malicious actors’ campaigns and how the use of AI systems has evolved over the past few months.

“Repeatedly, and across different types of operations, the threat actors we banned were building AI into their existing workflows, rather than building new workflows around AI,” states the document. “We found no evidence of new tactics or that our models provided threat actors with novel offensive capabilities.”

OpenAI highlighted several cases to demonstrate how threat actors use AI models. In one of the case studies, Russian-speaking cybercriminals attempted to develop malware — including features to evade detection, a remote-access trojan, and credential stealers — using ChatGPT. The chatbot detected the malicious intentions and refused to provide the code required, and OpenAI banned the user’s accounts.

In another case study, scammers from Cambodia, Nigeria, and Myanmar tried using ChatGPT for fraud. The malicious actors requested ChatGPT to write and spread messages across the internet to get potential victims’ attention using different approaches.

Researchers also noted that while hackers are using AI for malicious campaigns, users also rely on the technology to detect these activities.

“Our current estimate is that ChatGPT is being used to identify scams up to three times more often than it is being used for scams,” states the report. “As the threatscape evolves, we expect to see further adversarial adaptations and innovations, but we will also continue to build tools and models that can be used to benefit the defenders.”

OpenAI’s report has been released just a few weeks after cybersecurity experts warned about vulnerabilities in the company’s latest flagship model, GPT-5.