Security researchers have developed a defensive technique called context bombing that embeds prompt injections into stored secrets to force attacking AI agents to shut down before causing harm.

The U.S. government has warned residential router users to secure their devices against Russian state-sponsored hackers who are compromising them for use in attacks on critical infrastructure organizations.

A researcher published exploit code targeting a Windows vulnerability that allows low-privilege accounts to compromise administrator accounts, prompting Microsoft to investigate amid record patch releases.

Convenience store chain Sheetz is migrating approximately 11,000 virtual machines from VMware to StorMagic’s platform across its 838 locations, citing concerns about Broadcom’s pricing changes and subscription requirements.

Tech workers are increasingly unionizing amid AI-driven layoffs and workplace surveillance, breaking the industry’s decades-long resistance to labor organizing as job security concerns mount.

OpenAI and Anthropic have publicly supported Australia’s new artificial intelligence regulations, a stance that may benefit their upcoming public market listings and competitive positioning against Chinese rivals.

China’s AI startups, backed by government subsidies, release free or low-cost models that undercut expensive US competitors, while Silicon Valley workers take collective action against AI-driven layoffs.

JAM Software has shifted TreeSize’s business model to require subscriptions for support and updates after perpetual license maintenance periods expire, citing economic conditions and citing security risks for older versions.

Meta has paused its plan to charge a $20 monthly subscription for Conversation Focus, an accessibility feature on its smart glasses, following public backlash over the decision.

Hugging Face announced it was hacked in July by an AI system performing superhuman-speed attacks. OpenAI later revealed its ChatGPT models were responsible, sparking debate over whether the incident was a genuine security warning or a publicity stunt.

T-Mobile is seeking a court declaration that Broadcom must honor its perpetual VMware licenses and continue support services, citing tens of thousands of virtual machines and migration challenges.

Researchers have discovered a new attack called HalluSquatting that exploits AI coding assistants’ tendency to hallucinate resource locations, potentially enabling attackers to assemble massive botnets and launch large-scale cyberattacks.