The Claude maker paused model training for more than a month and disclosed testing failures that let AI models escape containment, all the while warning of doomsday scenarios.
The company resumes frontier model training offline after sandbox escapes and red-team tests exposed vulnerabilities in services that retained internet access.
AI surveillance towers watch millions along the US border, Former cops exploit Flock security gaps, DEA crypto detective unmasks dark web drug dealers, US-China AI arms control needed to curb cyberweapons, Tech firms urged to expose foreign disinformation tactics
Major US banks are restricting third-party AI access amid reports that some agents are attempting actions beyond customers’ instructions.
Two critical PHP flaws sent websites dark and defenders scrambling as disclosure unraveled—and evidence emerged of attacks days before patches arrived.
Emergency updates due today will address two critical vulnerabilities that researchers say could allow remote code execution across widely used PHP versions.
The ban also blocks future purchases and covers affiliates and third-party products built on DeepSeek technology.
The effort stalled, but election officials say the administration’s cuts have left CISA less equipped to defend voting systems against real threats.
Hacked cameras become weapons of war, China deepens its surveillance, Uganda slides into surveillance authoritarianism, Outlit wants AI agents to watch customers, Russian bots blackout is a myth Goodfire essay on neural networks is all the rage
Attack disrupts the platform serving every federal bankruptcy court, halting notice production and cutting access to key judiciary websites.
Citizen Lab uncovered an ad-delivered, zero-click spyware campaign targeting 45 Al Jazeera devices—no clicks or visible warning required.
Authorities dismantled more than 40 Texas laptop farms supporting fraudulent IT hires whose paychecks allegedly financed Pyongyang’s nuclear and missile programs.