Catch up on select AI news and developments from the past week or so (in no particular order):
Gemini adds Canvas document editing and Audio Overviews. Google has introduced two new features for Gemini: Canvas and Audio Overviews. Canvas allows users to draft, edit, and refine documents and code within Gemini, with integration to Google Docs for collaboration. The tool also supports code generation for apps and scripts with real-time previews. Audio Overviews, previously part of NotebookLM, are now available to summarize and present information via audio. Both tools are part of Gemini’s broader push into productivity and multimodal interfaces. Importance for marketers: These enhancements support AI-driven content creation and code development, offering marketers more efficient tools for ideation, copy generation, and automation workflows.
Study shows AI-generated meme captions score higher on average. A study comparing meme captions written by humans, AI, and human-AI teams found that captions by GPT-4o scored higher on average for humor, creativity, and shareability. AI consistently produced broadly appealing memes, though the funniest individual captions were still human-made. Human-AI teams created more ideas efficiently but felt less ownership. The study highlights AI’s role in boosting productivity while also noting emotional and qualitative trade-offs in creative collaboration. Importance for marketers: AI can increase output and audience resonance in content creation, but marketers should balance efficiency with human insight to maximize creative impact and emotional connection.
Stability AI unveils model to turn images into 3D videos. Stability AI released Stable Virtual Camera, a new model that converts up to 32 images into videos with simulated 3D depth and dynamic camera paths. The tool supports formats up to 1,000 frames and multiple aspect ratios. It’s currently available under a research-only, noncommercial license. Though early versions show limitations with complex textures and scenes, the model enables users to create immersive content without traditional 3D tools. Importance for marketers: This model presents new creative options for immersive advertising and storytelling, enabling brands to produce 3D-like video content from 2D assets at lower cost and complexity.
Mistral AI releases small model that outperforms GPT-4o Mini. Mistral AI has launched Mistral Small 3.1, an open-source model with 24 billion parameters that matches or exceeds GPT-4o Mini’s performance. The model processes text and images, supports 128k token contexts, and runs efficiently on modest hardware. Mistral is promoting digital sovereignty by making the model accessible under an Apache 2.0 license. The company also highlighted tools like Mistral OCR and Saba. Importance for marketers: Open, efficient AI models like Mistral Small 3.1 offer cost-effective alternatives for marketing applications, enabling on-device AI integration without reliance on large cloud-based proprietary platforms.
Baidu unveils two new AI models to rival US competitors. Baidu introduced two new AI models, including ERNIE X1, which it claims rivals DeepSeek R1 at half the cost. The model offers advanced reasoning and autonomous tool use. Baidu also launched ERNIE 4.5, with strong multimodal and high-EQ capabilities. These models aim to bolster Baidu’s position in the global AI race as Chinese firms compete with US tech giants. Importance for marketers: Competitive advances from global AI players mean more options and price competition for AI-driven tools, offering marketers new opportunities in content creation, personalization, and regional expansion.
Adobe launches agentic AI platform and prebuilt marketing agents. Adobe introduced agentic AI to its Experience Platform, offering 10 prebuilt agents for tasks like audience optimization, personalization, and content production. The new platform integrates with GenStudio Foundation and Brand Concierge for seamless campaign management. Adobe also announced data-sharing partnerships with AWS, Microsoft, SAP, and others to improve AI model training across platforms. The launch was announced during Adobe Summit in Las Vegas. Importance for marketers: Adobe's AI tools automate key marketing functions at scale. The interoperability and prebuilt agents help marketers streamline personalization, campaign planning, and content generation across ecosystems.
Anthropic reportedly working on voice capabilities for Claude. Anthropic plans to introduce voice interaction for its Claude AI assistant, according to company executive Mike Krieger. The feature is being prototyped for desktop use and could become a more natural interface for users. Anthropic has discussed potential partnerships with Amazon and ElevenLabs to support this capability. No launch date has been announced. Importance for marketers: Voice-enabled AI offers more intuitive user interaction. Marketers may explore voice-based engagement tools, conversational commerce, or hands-free content creation using Claude once this feature launches.
AI-powered search engines found to cite incorrect sources 60% of the time. A Columbia Journalism Review study revealed that generative AI search tools provided incorrect citations in over 60% of tests. Eight platforms, including ChatGPT and Perplexity, were evaluated using 1,600 queries. AI tools often generated plausible but incorrect answers rather than declining to respond. Some also ignored publisher restrictions. Paid versions frequently delivered more errors than free ones. Researchers raised concerns about growing reliance on these tools for news and research. Importance for marketers: Marketers relying on AI-driven search tools should vet information carefully. Misinformation risks can damage brand credibility and misguide strategic decisions based on flawed AI output.
Microsoft partners with Inait to develop brain-like AI models. Microsoft has partnered with Swiss startup Inait to create cognitive AI systems modeled on mammalian brain reasoning. The technology integrates symbolic reasoning with neural networks, aiming for adaptive, energy-efficient systems. Potential applications include finance, robotics, and enterprise tools. The partnership builds on 20 years of digital neuroscience research and will use Azure for deployment. Importance for marketers: Brain-inspired AI could yield more nuanced customer insights, smarter automation, and faster learning systems—advancing personalization and predictive marketing capabilities.
