New research from Anthropic shows how reward hacking in AI models can trigger more dangerous behaviors. When models learn to trick their reward systems, they can spontaneously drift into deception, sa [...]
Run a prompt injection attack against Claude Opus 4.6 in a constrained coding environment, and it fails every time, 0% success rate across 200 attempts, no safeguards needed. Move that same attack to [...]
MindsEye developer Build a Rocket Boy remains so convinced that corporate foul play contributed to the disastrous launch of its debut game that it’s now planning to prove it to its audience via in-g [...]
Model providers want to prove the security and robustness of their models, releasing system cards and conducting red-team exercises with each new release. But it can be difficult for enterprises to pa [...]
Dutch tech scaleup Optics11 has launched an underwater monitoring system that uses light waves to “listen” for the presence of foreign objects. Called OptiBarrier, the system can detect enemy subm [...]
WhatsApp has claimed that some users were “possibly compromised” by spyware, according to a report by The Guardian. The Meta-owned messaging app went on to allege that nearly 100 journalists and a [...]
The AI Gold Rush In 2025, we are in the midst of an AI arms race. Nearly every tech company, along with a growing number of businesses across virtually every sector – from finance, to healthcare, to [...]
Russia launched a covert operation to sabotage subsea cables while the world was distracted by the Middle East. [...]