Google Restricts New AI Model Over Safety Concerns

Published On 01 Oct, 2026
google-restricts-new-ai-model-over-safety-concerns

9 News HD: Google has taken a careful approach to the release of its latest AI model, limiting access while it studies possible safety risks linked to its advanced capabilities.

The company said the model will first be tested by selected experts, including cybersecurity professionals and the US government. Google plans to collect feedback from early users before deciding on a wider release.

The new AI system is built to handle difficult tasks in areas such as software development, finance, legal work and cybersecurity. Google says it has shown a strong ability to identify and fix serious software weaknesses.

During early testing, the model reportedly found a security problem in software used by hospitals worldwide. The issue could have exposed sensitive information, according to Google.

At the same time, cybersecurity experts are concerned that highly capable AI could be misused to target banks, hospitals and government networks. Google says the model has safety controls to reject requests that could assist in cyberattacks or the development of dangerous weapons.

The company is also watching how the model reasons to reduce the risk of it acting beyond what users intended. Google says a phased release will allow it to better understand these risks before making the AI system available to a wider audience.