10 stories
On September 24, 2026, the Blueprint Alliance, led by Okta, announced a proposal for an AI agent kill switch to secure businesses against rogue AI activity.
OpenAI's Daybreak rollout to Ukraine highlights the potential of AI-driven cyber defense, but also raises concerns about model safety and the need.
Microsoft disrupts EvilTokens, an AI-assisted fraud service that compromised 12,000 accounts, detailing the takedown, incentives, and next-step defenses.
The Claude Opus 5 hack OpenAI incident shows how off-the-shelf LLMs can generate functional exploits, highlighting urgent needs for supply-chain hygiene.
Claude token theft reveals critical gaps in Anthropic's usage monitoring and malware defenses, urging developers to tighten credential hygiene and demand.
OpenAI $1 billion Daybreak program subsidizes AI tools for utilities, local governments and other essential-service defenders.
Google Fairwind Program offers governments and select enterprises cutting-edge AI-driven cyber-defense tools, detailing architecture, hardware.
Google's HEIR toolkit brings homomorphic encrypted inference to the forefront, allowing developers to deploy AI models with end-to-end encryption.
Google fixed over 1,072 Chrome bugs in June with AI-powered tools, enhancing Chrome security updates and setting a new record.
OpenAI introduces Lockdown Mode to protect sensitive data from prompt injection attacks with enhanced security features and reduced risks.