Cybersecurity
Cybersecurity
Anthropic security executive warns AI-found software flaws will get worse for 18 months
On 22 October 2026, Jason Clinton, security chief at Anthropic, posted a blunt warning: "as bad as things have gotten in the past 4 days, they are going to get a lot worse before they get better. security leaders, never let a crisis go to waste: please go to your boards and ask for help right now.
Cybersecurity
Hundreds of thousands of servers run PHP versions that will never get the PHP2Shell fix
Small businesses, local councils, sports clubs and community charities running cheap shared hosting carry the heaviest exposure from two critical PHP vulnerabilities disclosed on 21 October 2026.
Cybersecurity
Patching your web server won't stop PHP2Shell in your firewall
PHP2Shell reaches well past web servers. The two critical remote code execution vulnerabilities, tracked as CVE-2026-109618 and CVE-2026-109619, live inside any PHP-enabled interface, including the management consoles of enterprise firewalls, NAS storage devices and network appliances that embed
Cybersecurity
Governments pull websites offline as hackers target critical PHP flaws
Attackers are already trying to take over websites through two critical PHP flaws. WordPress updates won't protect you; your host has to act, today.
Technology
Spyware-linked ads land on Australia's nuclear agency networks
The same advertising pipeline infected Al Jazeera devices without a click. Australia's nuclear science agency says it has found no impact on its systems.
Cybersecurity
JPMorgan boss warns AI multiplies global cyber risk tenfold
Researchers say Anthropic's published claims lack independent reproduction, while financial services firm FIS uses the restricted model for defence.
Cybersecurity
OpenAI ignored warnings before AI agents accessed Medicare portal
OpenAI notified Australia nearly three months after the incident. The company says no individual patient records were accessed.
Cybersecurity
OpenAI agent read Medicare files 98 days before Australians knew
An OpenAI agent read non-public files and wrote data on a Services Australia server, and Australians were told 98 days later. Jane Hume called the agency's cyber defences woefully inadequate, and the case reads as a misconfiguration story as much as an AI one.
Cybersecurity
Chinese AI labs route customer chats to Claude without consent
Alibaba extracted 151 million exchanges to train its Qwen models, while other operations exposed live Russian defence credentials and surveillance footage linked to the People's Liberation Army.