/i.s3.glbimg.com/v1/AUTH_59edd422c0c84a879bd37670ae4f538a/internal_photos/bs/2021/H/w/YbA657S3aYVfC0P9wboQ/g1-favicon.png){kind=small}
/i.s3.glbimg.com/v1/AUTH_da025474c0c44edd99332dddb09cabe8/internal_photos/bs/2023/1/X/nkkB7tSdirnIUbGhakCQ/favicon-o-globo.png){kind=small}
/i.s3.glbimg.com/v1/AUTH_63b422c2caee4269b8b34177e8876b93/internal_photos/bs/2018/e/e/PAMa3oQXOPBUOsGgFyFw/valor-tenant-icon.png){kind=icon}
Overview
- Google says it disrupted an intrusion that used a chatbot‑style model to locate a zero‑day bug, a flaw with no patch, that let the attackers skip two‑factor login and reach a system admin tool.
- The company reported no sign of government involvement and said it alerted the targeted firm and law enforcement after cutting off the operation.
- A national survey in Brazil found 18% of health providers used AI in 2025, with 76% of adopters relying on chatbots such as ChatGPT and Gemini and larger hospitals showing the highest uptake.
- Security studies and experts warned that unplanned AI rollouts raise data exposure and operational risks, with Thales reporting about 70% of organizations now see AI as a top data‑security threat.
- Defenders are also turning to AI, with the Pentagon deploying Anthropic’s Mythos to find and fix software flaws, while recent cases in Mexico showed criminals jailbreaking models to automate large‑scale data theft.