Gemini

PilotCite helps brands get cited by ChatGPT, Gemini, and more

PilotCite is a platform that enables brands to be cited by AI models like ChatGPT and Gemini. It aims to increase brand visibility within AI-generated content.

Why it matters: As AI models increasingly generate content, being cited by them can enhance brand recognition and credibility. PilotCite offers a way for brands to engage with this emerging channel.

Alphabet delays rollout of Gemini 3.5 Pro AI model

Alphabet announced its Gemini 3.5 Pro AI model in May, initially planning a broader rollout the following month. However, the release has been delayed despite internal use of the model.

Why it matters: Delays in deploying advanced AI models like Gemini 3.5 Pro can impact Alphabet's competitive position in the AI market. Investors reacted negatively, reflecting concerns about the timeline for new AI capabilities.

Google Vids adds personalized AI avatars for self-starring videos

Google Vids now allows users to create videos featuring digital avatars of themselves using Gemini Omni-powered tools for video generation and editing. These tools enable video creation from prompts and reference images.

Why it matters: This update enhances user engagement by enabling personalized video content creation with AI avatars, simplifying the video production process. It reflects ongoing advancements in AI-driven multimedia tools.

Reliability of Gemini Models as Audio Judges for Full-Duplex Voice Agents

The study evaluates the reliability of Gemini models (2.5 Flash, 3.5 Flash, 3.1 Pro) as audio judges scoring full-duplex voice agent conversations from raw stereo waveforms. Gemini 2.5 Flash was validated against human raters across 209 sessions on eight production dimensions.

Why it matters: Reliable automated scoring of voice agent conversations can improve evaluation efficiency and consistency compared to human raters. This supports development and benchmarking of full-duplex voice agents using scalable, objective metrics.

Startup Tackles Groupthink in Large Language Models

Large language models like Claude, ChatGPT, and Gemini often produce similar outputs, exemplified by their tendency to pick the number 7 when asked for a random number between 1 and 10. A startup is working on methods to diversify these models' responses and break this pattern of groupthink.

Why it matters: Reducing groupthink in LLMs can lead to more varied and creative AI outputs, enhancing their usefulness and reliability. This effort addresses a fundamental limitation in current consumer AI models.

Unlocking dependable responses with Gemini Enterprise Agent Platform’s Agentic RAG

Data Management

Google Workspace Tackles Indirect Prompt Injection Threats

Google's GenAI Security Team outlines a continuous approach to mitigate indirect prompt injection (IPI), an evolving threat vector targeting AI systems. The strategy focuses on ongoing detection and response to protect Google Workspace users.

Why it matters: As AI systems become more integrated into workflows, vulnerabilities like IPI pose significant security risks. Google's proactive measures highlight the importance of adaptive defenses in AI security.

Architecting Security for Agentic Capabilities in Chrome

Posted by Nathan Parker, Chrome security team Chrome has been advancing the web’s security for well over 15 years, and we’re committed to meeting new challenges and opportunities with AI. Billions of people trust Chrome...

LLMs and GenAI Pose Growing Risks to Cybersecurity and Platform Integrity

Large Language Models (LLMs) and generative AI systems like ChatGPT and Gemini are transforming digital platforms but also raising significant cybersecurity, privacy, and platform integrity challenges. Notably, LLM-assisted malware is projected to increase from 2% in 2021, highlighting escalating r...

Why it matters: As LLMs become more integrated into various sectors, their misuse in malware and scams threatens digital trust and safety. Understanding these risks is crucial for developing effective safeguards in AI deployment.