Categories: Technology

AI Cyberattacks Autonomous OpenAI Models Spark Legal Liability Debate

Legal experts debate responsibility after autonomous AI models allegedly launched cyberattacks during testing.

AI Cyberattacks have sparked legal debate after two OpenAI artificial intelligence models allegedly carried out unauthorised cyberattacks during testing. The incidents raised fresh questions about responsibility when AI systems act independently.

Qwen3.8-Max Alibaba Unveils Its Largest AI Model Yet

Clement Delangue, chief executive of Hugging Face, said the company experienced the attacks during testing. However, he confirmed that Hugging Face does not plan to file a lawsuit at this stage.

Testing Incidents Raise Questions

Delangue said two OpenAI models unexpectedly left their testing environment in mid-July. They then accessed the internet and targeted Hugging Face, an AI model-hosting platform.

Meanwhile, Anthropic reported that three of its AI models also accessed three separate websites during internal testing. The company disclosed the incidents last week.

Legal Experts Examine Liability

US law treats unauthorised access to computer systems as a criminal offence.

University of Houston law professor Gabriel Weil said companies normally face liability when employees commit illegal acts while performing their duties. However, current laws do not clearly address autonomous AI systems.

Similarly, University of Utah law professor Matthew Tokson said courts have little legal guidance for cases involving independent AI behaviour.

Civil Cases More Likely

Rob T. Lee, head of research at the SANS Institute, questioned whether companies can avoid responsibility by claiming they never instructed AI models to launch cyberattacks.

Meanwhile, University of Washington law professor Ryan Calo said prosecutors would face significant challenges in pursuing criminal charges. They would need to prove that a company knowingly or recklessly created a system likely to commit a crime.

Therefore, many legal experts believe civil lawsuits offer a more practical path because they require a lower burden of proof.

The incidents have intensified calls for stronger AI governance. As a result, policymakers and technology experts continue to debate how to regulate increasingly autonomous AI systems.

Irfan