Shared API keys expose AI agents at 69% of enterprises, new VentureBeat research finds
THE SO WHAT
Treating AI agents like scripts and reusing API keys means a single compromised agent inherits the blast radius of every workflow that key touches. This week, inventory shared credentials across your agent fleet and move to per-agent or per-workflow keys with auditable scopes.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIOpenAI launches its new family of models with GPT-5.6
A new GPT‑5.6 family with explicit cybersecurity upgrades means “model choice” is now a security decision, not just a quality decision. CISOs and CTOs should be in the loop on model selection and threat modeling, not just the AI team.
Applied AIInstagram’s AI image generator alarms privacy experts
Meta training an AI image maker to default on public profile photos is another step toward “public by design” identity surfaces. Consumer apps are normalizing synthetic versions of you—enterprises need to assume employees’ likenesses and brands will be remixed in ways no policy has yet anticipated.
Applied AIAn AI agent startup just let its agent run its $100 million fundraise
Letting an AI agent run a $100 million raise is a marketing stunt and a stress test—capital formation is now a reference workflow for agentic systems. If you’re selling agents into enterprises, you now have a public “closed the round” case study to point to, but also a higher bar for auditability and compliance.
Applied AIOpenAI is shutting down Atlas, but its AI browser ambitions are still growing
Killing Atlas while moving agentic browsing into the desktop app and Chrome extension shows the browser is a feature, not a destination. If your product depends on “AI inside the tab,” expect those workflows to be absorbed into assistants that live above or alongside the browser, not as separate apps.