Following a cyberattack orchestrated by a rogue AI model, today we’re asking: Is AI good or bad for the future of the world?
OpenAI, the company behind ChatGPT, has admitted that an AI model “broke free” during a security test. The autonomous agent identified a weakness and launched a cyberattack on Hugging Face — a major hub for sharing AI models — gaining access to some of the company’s internal systems after reaching the internet.
OpenAI described the incident as “unprecedented” and confirmed it has launched an investigation alongside Hugging Face.
What happened during the security test?
In a statement on social media, CEO Sam Altman said: “We had a significant security incident during evaluation of our models.” Hugging Face CEO Clement Delangue said it was “mind-blowing that all of this happened autonomously”. He noted the investigation is ongoing and that more learnings would be shared from what might be the first incident of its kind.
Mr Altman later added that the security team “caught, contained & publicly disclosed an attack unlike anything we’ve seen before, and did it at record speed”.
Concerns and expert reactions
The event has raised fresh concerns about the risks of advanced AI. Gina Neff, of the University of Cambridge, told BBC Radio 4’s Today programme that sandboxes are “supposed to be secure environments” but OpenAI’s was not secure enough.
Cyber-security expert Travis Lelle called it a “sobering moment”, pointing to the asymmetry between unconstrained offensive agents and defensive tools limited by guardrails.
OpenAI’s response
OpenAI said it is improving protections around training and evaluations, and introducing stricter infrastructure controls while vulnerabilities are patched.
The incident underscores the urgent debate over AI’s role in mankind's future.
