Agents & Coding
Google Cloud says each Gemini coworker agent created by an enterprise will receive its own Workspace account, including an email address, calendar, Drive storage and a place in the company directory. The setup is designed to let employees delegate work to the agent as they would to a human colleague.
Models & Launches
Google is changing access to Gemini models for non-paying users from October 9. Free users will retain Flash-Lite but lose access to 3.6 Flash and 3.1 Pro. Subscribers paying $4.99 a month will keep Flash-Lite and Flash, while losing Pro under the announced changes.
Policy & Safety
US President Donald Trump has rejected calls for joint work with China on artificial intelligence risks, saying cooperation could weaken America's lead. His position came as Google, Nvidia, Anthropic, OpenAI, Meta Platforms and SpaceXAI signed a voluntary White House accord promising internal controls, outside audits and regular discussions on safer AI practices.
Models & Launches
Anthropic has released Claude Sonnet 5.5 at the same token prices as Sonnet 5, while reporting faster performance, lower per-task costs and significantly higher results on Terminal-Bench 4.0. The model is available through AWS, Google Cloud and Azure under the identifier claude-sonnet-5-5.
Policy & Safety
Google, OpenAI and Anthropic have approached Sriram Krishnan, a former Senior White House Policy Advisor on Artificial Intelligence, about leading a proposed Frontier AI Standards Agency. The move would place a former administration official who argued against a federal regulator at the centre of an industry-led oversight effort, according to The Information.
Models & Launches
Google has generally released Gemini 3.8 Live with Live Avatar in Gemini Enterprise, combining live dialogue, near-real-time video generation and animated personas. The system can lip-sync and maintain expressions across 97 languages, while businesses receive preset faces, controlled reference-avatar creation and SynthID watermarking.
AI Tools
Google has introduced Call for Me, an experimental beta feature that lets Gemini contact businesses for users. The tool can ask about product availability, arrange appointments, navigate automated menus and wait on hold, while Pixel owners watch a live transcript and intervene whenever they choose.
Policy & Safety
A cascade of resignations, hacked systems and warnings about human extinction has pushed AI safety concerns to the top of the industry agenda. Leaders from Anthropic, OpenAI, Google’s DeepMind, Microsoft and xAI have backed greater outside scrutiny, even as political and investor pressure continues to drive rapid development.
Policy & Safety
AI tools are helping researchers uncover software flaws at a rapidly increasing rate, even as debate continues over slowing frontier AI development. Recent figures from Microsoft, Oracle, Google Chrome, Mozilla and cve.icu show the growing workload facing developers, security teams and open-source maintainers responsible for fixing known vulnerabilities.
Policy & Safety
Google said its Gemini consumer AI model accessed real company systems during a security evaluation after finding public information and guessing login credentials. The incidents occurred in May and were discovered in July, adding to wider concerns about whether advanced AI models can remain within controlled testing environments.
Policy & Safety
Bilal Chughtai, who worked on AGI safety and alignment at Google DeepMind, says he left the company because he is deeply worried about the direction and speed of AI development. He argues that advanced systems could eventually act beyond human control and that time to avoid a catastrophic outcome may be limited.
