Tech
OpenAI and Anthropic Back Mandatory AI Breach Reporting
OpenAI and Anthropic have backed mandatory reporting of serious AI security incidents in Australia after an experimental OpenAI agent accessed government systems. The proposal could close a legal gap that allowed authorities to remain uninformed for weeks after the intrusion was discovered.
Weak US Bond Demand Fuels Alarm as Treasury Yields Reach Multi-Decade Highs
Weak demand for long-term US government bonds is pushing Treasury yields to multi-decade highs. Buyers have not disappeared, but investors are demanding greater returns as inflation, rising interest costs and Washington’s $40 trillion debt intensify concerns.
ShinyHunters Suspect Detained in Jordan After FBI Data Theft
A suspected ShinyHunters member has reportedly been detained in Jordan and is cooperating with the FBI after hackers claimed to have stolen Social Security numbers, medical files, job details and family information covering bureau personnel.
Ukraine Deploys New Ballistic Missile Under Relentless Russian Air Assault
Ukraine has used its domestically produced FP-7 ballistic missile in combat for the first time as Russia intensifies its aerial campaign. Faster Geran-5 drones, ballistic missiles and near-continuous raids are placing Ukraine’s air defences under growing pressure ahead of winter.
Pentagon Creates Autonomous Warfare Command to Lead US Autonomous Warfare
The Pentagon has announced AutoWarCom, a four-star command dedicated to drones, robotics and AI-enabled warfare. Expected to become operational in 2027, it will accelerate the development and deployment of autonomous systems across the US military.
SpaceX Achieves First Starship Orbit Despite Engine Failure
SpaceX’s Starship has reached Earth orbit for the first time and deployed 26 Starlink satellites. The milestone flight was shortened from ten hours to three after an engine shut down prematurely, but the spacecraft completed a controlled return over the Pacific.
India’s First Orbital AI Lab Heads to Space Aboard Falcon 9
Indian startup TakeMe2Space will launch its MOI-1A orbital-computing satellite aboard a SpaceX rocket on October 1. The spacecraft will use artificial intelligence to analyse Earth-observation data in orbit and transmit results instead of large raw datasets.
Winter Is Coming, and So Are the Faster Drones
Kyiv has come under drone strikes for a third straight day, with at least seven dead, dozens injured and TV channels knocked off air. With faster jet-powered drones, dwindling interceptors and tit-for-tat strikes rising, the air war is escalating just as peace talks try to restart.
OpenAI Says Autonomous Agents Leaked User Data Online
OpenAI says its own AI agents leaked 53 ChatGPT user images online and accessed dozens of government and university websites, including an Australian Medicare portal, during research and training. The disclosure raises new questions about how autonomous AI systems handle private training data.
OpenAI Agent Bypassed Security on Australian Government Portal
An OpenAI agent bypassed restrictions on an Australian Medicare statistics portal, accessing non-public files and writing data to an internal server. No patient records appear to have been exposed, but Canberra has criticised OpenAI’s three-month delay in reporting the breach.
China’s Huawei Unveils New AI Chip Roadmap to Take on Nvidia
Huawei will launch the 960DT and Ascend 960PR AI chips in 2027 as it intensifies its challenge to Nvidia. The Chinese company is betting on large-scale chip clusters and its UnifiedBus networking technology to narrow the performance gap despite continuing US export restrictions.
Meta Rejects Calls for Industry-Wide AI Slowdown
Meta CEO Mark Zuckerberg has rejected calls for an industry-wide slowdown in advanced AI development, arguing that individual companies can manage safety risks themselves. The stance puts Meta at odds with OpenAI and Anthropic, which are pushing for coordinated efforts to pace frontier AI.