Catch up on select AI news and developments from the past week or so (in no particular order):
Google quietly releases Gemini 2.0 Pro Experimental in chatbot changelog. Google has introduced Gemini 2.0 Pro Experimental, its newest flagship AI model, through a low-profile changelog update rather than a major announcement. Positioned as the successor to Gemini 1.5 Pro, the model is aimed at enhancing coding and mathematical capabilities. However, its release was later removed from the changelog, with Google calling it an “out-of-date release note.” Importance for marketers: The discreet launch strategy suggests Google is adopting an iterative approach to AI releases, emphasizing rapid experimentation. Marketers should monitor developments in Gemini AI as Google refines its capabilities for business applications.
Mistral AI unveils small, efficient model to rival big tech’s AI. Mistral AI has launched Mistral Small 3, a compact, high-performing AI model that rivals much larger models in accuracy and efficiency. With 24 billion parameters, the model achieves 81% benchmark accuracy and is released under an open-source Apache 2.0 license. Mistral’s approach prioritizes training optimization over sheer scale, positioning the model for enterprises needing on-premises AI solutions. Importance for marketers: The rise of smaller, cost-effective AI models could drive broader adoption in marketing applications, reducing reliance on expensive cloud-based AI solutions while improving data privacy and operational efficiency.
US copyright office allows AI-assisted works to receive copyright with human creativity. The US Copyright Office has clarified its stance on AI-assisted works, stating that they can receive copyright protection if they contain significant human creativity. A new report emphasizes that fully AI-generated content remains ineligible, but human modifications to AI-generated outputs could qualify for protection. The office is also preparing a separate report on AI model training and copyright concerns. Importance for marketers: This decision sets a precedent for creative professionals and brands using AI tools, reinforcing the need for human oversight in AI-generated content to secure intellectual property rights.
Alibaba unveils Qwen 2.5 AI model, claims it surpasses DeepSeek. Alibaba has launched Qwen 2.5-Max, claiming it outperforms leading AI models, including DeepSeek-V3, GPT-4o, and Meta’s Llama 3.1. The move follows intense competition in China’s AI sector, where price wars and rapid advancements have reshaped the market. (See more on DeepSeek, below.) Importance for marketers: The increasing rivalry among Chinese AI firms highlights the global race for AI dominance, potentially impacting AI service costs and availability for businesses worldwide.
Convergence’s Proxy AI assistant outperforms OpenAI and Anthropic in benchmarks. Convergence AI has launched Proxy, a highly customizable AI assistant that automates repetitive digital tasks. Proxy surpasses competitors in key performance benchmarks, offers automation capabilities, and includes a global rollout with competitive pricing. The company claims Proxy excels in task memory and execution compared to OpenAI’s Operator. Upcoming features include parallel agent multitasking and custom workflows. Importance for marketers: Proxy’s automation potential could significantly improve productivity by streamlining workflows, making AI-driven task management more accessible and affordable for marketing teams.
California AG warns AI companies of potential legal violations. California Attorney General Rob Bonta has issued legal advisories outlining how AI companies may be violating state laws. Key concerns include AI-generated deception, false advertising of AI capabilities, and potential discrimination in automated decision-making. The advisory signals increased regulatory scrutiny of AI business practices. Importance for marketers: This legal stance raises compliance risks for AI-driven marketing campaigns, emphasizing the need for transparency in AI-generated content, ad targeting, and automated decision-making.
Charity launches AI fact-checking agents as Meta disbands program. The Society Library has introduced AI-powered fact-checking agents, @MuseofTruth and @MuseofResearch, on social platforms. These agents use a proprietary system to analyze and verify information, offering an alternative to Meta’s discontinued fact-checking program. The initiative is backed by cryptocurrency donations. Importance for marketers: As misinformation concerns grow, AI-driven fact-checking tools could help brands maintain credibility and counteract false narratives in digital spaces.
OpenAI launches ChatGPT Gov for US government agencies. OpenAI has released ChatGPT Gov, a specialized AI version for US government agencies, integrated with Microsoft Azure’s cloud services. The launch follows increased competition from Chinese AI startup DeepSeek, which has disrupted the market with a free AI assistant. Importance for marketers: Government adoption of AI signals mainstream acceptance, potentially leading to regulatory developments and expanded use cases in public-sector communications and services.
Apple revamps AI team and plans major Siri upgrades. Apple has reassigned longtime executive Kim Vorrath to its AI division as a top deputy, signaling a push to improve its AI initiatives. Siri is expected to receive a major upgrade in iOS 18.4, enabling better on-screen comprehension. Apple also paused its AI-powered news summary feature due to accuracy concerns. Importance for marketers: Apple’s renewed focus on AI suggests future improvements in voice search and virtual assistant capabilities, potentially influencing marketing strategies for voice-driven search and customer engagement.
Samsung adds AI customization for Sketch to Image in Galaxy S25. Samsung’s One UI 7 introduces a new feature for Galaxy S25 users, allowing them to adjust AI involvement in the Sketch to Image tool. A “Sketch matching strength” slider lets users control how much AI alters their drawings. Importance for marketers: Customizable AI-generated visuals could benefit creative teams, offering more control over AI-assisted design outputs in marketing materials.
Apple working on smart glasses and a Meta Ray-Ban rival. Apple is reportedly developing multiple smart glasses concepts, including an XR-compatible version of visionOS and a competitor to Meta’s Ray-Ban smart glasses. The company is refining its wearable AI strategy following the lukewarm reception of the Vision Pro headset. Importance for marketers: Smart glasses could open new avenues for immersive marketing, AR-powered advertising, and hands-free digital interactions.
Google invests $120M in AI education amid regulatory scrutiny. Google is expanding AI education programs globally, investing $120 million to train workers and influence AI policy discussions. The move follows regulatory challenges in advertising and search. Google is partnering with colleges to offer AI-related job training. Importance for marketers: AI workforce development could shape future marketing talent pools, increasing AI fluency among professionals.
OpenAI faces new copyright lawsuit from Indian publishers. Indian and global book publishers have sued OpenAI, alleging that ChatGPT generates unauthorized book summaries. The case follows similar lawsuits worldwide, raising questions about AI’s use of copyrighted content. Importance for marketers: This lawsuit underscores the growing legal risks surrounding AI-generated content, impacting how brands use AI for content creation and summarization.
All About DeepSeek This Week
What is DeepSeek, and why is it disrupting the AI sector?. DeepSeek’s AI models, developed at a fraction of the cost of competitors, are challenging the dominance of US tech giants. The company’s open-source approach and cost-efficient AI models have led to rapid adoption, raising concerns about the sustainability of billion-dollar AI investments. Importance for marketers: The disruption caused by DeepSeek highlights the growing accessibility of high-performance AI, which could lower costs and expand AI applications in marketing.
DeepSeek unveils Janus Pro, an open-source rival to DALL-E 3. DeepSeek has released Janus Pro, a new open-source multimodal AI model that reportedly outperforms OpenAI’s DALL-E 3 in key benchmarks. The model supports text and image generation but requires fine-tuning to match specialized models. Unlike US tech firms, DeepSeek continues to open-source its releases. Importance for marketers: Open-source AI models could democratize access to advanced AI tools, allowing marketers to develop custom AI-driven creative content at lower costs.
DeepSeek limits registrations after large-scale cyberattack. Chinese AI startup DeepSeek announced it is temporarily restricting new user registrations due to malicious cyberattacks. The company recently gained prominence after overtaking OpenAI’s ChatGPT as the most downloaded app on Apple’s App Store. DeepSeek’s rapid rise has sparked debate over the sustainability of billion-dollar AI investments. Importance for marketers: The attack highlights security risks in the AI industry, reinforcing the need for robust cybersecurity measures as AI adoption accelerates.
DeepSeek surpasses ChatGPT as the most downloaded app. DeepSeek’s AI Assistant app has taken the top spot on Apple’s App Store in 51 countries, surpassing ChatGPT. The rapid adoption follows the release of its open AI models, which claim significant efficiency gains in training costs. The surge in popularity has unsettled the AI industry, prompting scrutiny of DeepSeek’s training methods. Importance for marketers: DeepSeek’s rise signals shifting consumer preferences in AI tools, emphasizing the need for companies to track emerging competitors in the generative AI market.
Researchers recreate DeepSeek’s core AI for just $30. University of California, Berkeley researchers claim they replicated DeepSeek’s reinforcement learning technology for just $30, raising questions about the high costs associated with AI development. Their model, though smaller, demonstrated iterative problem-solving and accuracy improvements. Experts remain skeptical of DeepSeek’s reported low training costs, warning of hidden expenses. Importance for marketers: If accurate, these findings suggest cost-efficient AI models may soon emerge, reshaping AI adoption and accessibility for businesses.
OpenAI investigates whether DeepSeek copied its models. OpenAI is reviewing allegations that DeepSeek used its models to develop competing AI systems. US officials warn that Chinese firms may be using “distillation” techniques to extract knowledge from US AI models. The dispute comes amid growing tensions over AI leadership between the US and China. Importance for marketers: Intellectual property concerns in AI could impact cross-border collaborations and regulatory decisions, shaping how AI technologies are developed and commercialized.
DeepSeek chatbot scores low in accuracy compared to rivals. A NewsGuard audit found that DeepSeek’s chatbot achieved just 17% accuracy in delivering reliable news information, ranking 10th out of 11 AI models tested. The model repeated false claims 30% of the time and defaulted to vague answers in 53% of cases. Importance for marketers: Accuracy issues in AI-generated content underscore the need for brands to vet AI outputs carefully to avoid misinformation in marketing and communications.
DeepSeek founder Liang Wenfeng emerges as key figure in AI race. Liang Wenfeng, founder of DeepSeek, has rapidly gained prominence as a leading figure in China’s AI industry. He advocates for innovation over imitation, positioning DeepSeek as a research-driven AI company. DeepSeek’s success has drawn attention from Chinese policymakers, signaling potential government backing. Importance for marketers: China’s AI push could reshape global AI competition, influencing market dynamics and investment priorities in AI-driven marketing technologies.
You can find the previous issue of AI Update here.
Editor’s
note: GPT-4o was used to help compile this week’s AI
Update.
