Citizen Lab has found evidence of governments using online ads to spread 'zero-click' spyware, which can hack devices without the user's knowledge or interaction. The digital rights unit discovered that dozens of devices at Al Jazeera were compromised after being shown malicious online ads. The exact number of devices affected and the specific governments involved are not specified.

AI agents are not becoming 'rogue' as often as feared, but researchers are discovering that many AI labs are skipping fundamental security basics, leaving them vulnerable to attacks. This is because researchers are often focused on the technical aspects of developing AI models, rather than the security implications. As a result, AI models are being deployed without adequate security measures in place, making them susceptible to attacks. This is a concern, as AI models are increasingly being used in critical systems, such as healthcare and finance, where security breaches could have serious consequences.