AI Image
The PHP Meltdown Shows How AI Makes Bugs More Exploitable
WIRED • • general
Key Points:
- A recently patched vulnerability in ChatGPT’s Mac app exposed sensitive user data, highlighting that AI software itself can be a target for hackers.
- Two critical PHP vulnerabilities, known as PHP2Shell bugs, have been disclosed and are being widely exploited, with Amazon Web Services detecting attacks days before patches were released.
- AI is increasingly used in anti-cybercrime efforts by deploying lifelike bots to deceive cybercriminals, though hackers continue to leverage agentic AI tools to conduct sophisticated attacks.
- Meta has suspended a feature in its Muse AI assistant after a design flaw in share links was found to expose users’ personal conversations to unauthorized access.
- OpenAI delayed the release of its latest Astra model due to safety concerns and apologized for a prior security breach involving an Australian government website.
