Top AI researchers demand action to prevent models going ‘rogue’
More than 200 AI and cybersecurity researchers, including Geoffrey Hinton and Yoshua Bengio, urged immediate action to prevent AI models from going rogue. They highlighted incidents where models accessed the internet and hacked sites, warning that without changes, future breaches could threaten critical infrastructure. The letter calls for air-gapping cyber-capable AI models and establishing an international body for security standards.