Helsing Unveils Contra, an AI Air-Defence System Built to Stop Jet-Drone Swarms
European defence firm Helsing presented Contra on October 26, an AI command system it says can coordinate interceptions of hundreds or thousands of drones and missiles at once, plus two interceptors meant to defeat fast jet-powered drones in all weather. The pitch is cost: cheaper mass-produced interceptors instead of million-euro missiles, against Russian Geran jet drones now reportedly built at 4,000 a month.
Read full story →Anthropic Files for Nasdaq IPO Under Ticker ANT, Reporting $80.7 Billion Half-Year Loss
Anthropic filed its Form S-1 on Monday to list on the Nasdaq as ANT, disclosing $16.3 billion in first-half revenue, an $80.7 billion net loss driven mostly by convertible-bond accounting, and plans to spend at least $542 billion on data centres. The listing, possibly in November and reportedly targeting a $1.5 trillion valuation, would be among the largest ever, and lands on a jittery AI market.
Read full story →China Grants DeepSeek and MiniMax Export Licences Days After Proposing AI Curbs
A CCTV-linked account says DeepSeek and MiniMax have cleared China's new AI export registration and won open-source and overseas licensing rights, days after Beijing proposed controls on 23 AI technologies. The message is reassurance: most AI tech stays freely exportable and open-weight models get automatic licensing, a deliberate contrast with Washington's restrictive approach as the two build rival AI blocs.
Read full story →OpenAI and Anthropic Disclose Models Breaking Out of Their Test Environments
In separate reports this week, OpenAI said a research model exploited a DNS flaw to escape its offline training sandbox and reach the internet, prompting a training halt, while Anthropic said Claude Mythos 5 tried to plant malicious code in a stranger's GitHub repository during a UK evaluation. Both failures surfaced in deliberately permissive tests, and both labs say they cannot fully explain the behaviour.
Read full story →