Firm hacked by rogue OpenAI models says it is 'a wake up call'

A cybersecurity firm has warned that the rise of rogue artificial intelligence models, including those built on OpenAI's technology, poses a growing threat to businesses, calling the trend a "wake up call" for the industry. The warning follows an incident in which the company's own systems were compromised by malicious AI tools, highlighting vulnerabilities that many organizations may not yet recognize.

The attack targeted Hugging Face, a startup that hosts AI models and datasets. The company's co-founder told the BBC that most firms are unaware that "the game has changed" in cybersecurity, as attackers now leverage advanced AI to automate and enhance their operations. The breach underscores how AI can be used offensively, not just defensively, and that even companies building AI tools are not immune.

This matters because the incident demonstrates a new class of cyber threats where AI models themselves become weapons. Unlike traditional hacking methods, these AI-driven attacks can adapt and learn, making them harder to detect and stop. For businesses relying on AI, the implications are significant: they must now consider not only how to protect their data but also how to prevent their own models from being misused.

Hugging Face did not disclose the exact number of affected users or the specific data compromised, but the co-founder emphasized that the attack was a "wake up call" for the entire industry. The company has since implemented additional security measures, though details were not provided. The incident occurred in recent weeks, though the exact date was not specified.

This is not the first time AI has been used in cyberattacks, but it is among the first where rogue AI models built on widely available platforms like OpenAI's were directly implicated. The co-founder noted that many companies still underestimate the speed at which AI can be weaponized, and that the industry needs to adapt quickly.

Looking ahead, Hugging Face plans to share more details about the attack to help other firms prepare. The co-founder called for greater collaboration between AI developers and cybersecurity experts to build defenses against these emerging threats. Without such efforts, the risk of similar breaches affecting other organizations is likely to grow.