Google launches Gemini 2.0 AI Models + ByteDance showcases hyperrealistic deepfakes

Today’s Highlights:

?? News: Google launches Gemini 2.0 AI Models + ByteDance showcases hyperrealistic deepfakes

?? Funding: A EU invests $56M in open-source multilingual AI model compatible w/ all EU languages

?? Top News Stories:

1. 谷歌 launches Gemini 2.0 AI Models:?

• Gemini 2.0 Pro excels in coding, complex prompts, and reasoning, featuring a massive 2 million-token context window—allowing it to process the equivalent of all seven Harry Potter books in one prompt.
• Gemini 2.0 Flash is now generally available, this model is optimized for high-efficiency AI applications, offering low-latency responses and supporting multimodal reasoning with a context window of one million tokens.
• Gemini 2.0 Flash-Lite is designed as a cost-effective solution, Flash-Lite surpasses the previous 1.5 Flash model in most benchmarks while maintaining the same speed and price. It supports multimodal input and features a 1M token context window.
• The new Gemini 2.0 models are available starting Wednesday via Vertex AI, Google AI Studio, and Gemini Advanced subscriptions.

https://youtu.be/Fs0t6SdODd8?si=MC0JMxxnQ-hPIsdW?

2. 字节跳动 researchers unveiled OmniHuman-1, an AI system that creates hyper-realistic deepfake videos from a single image and audio clip, raising concerns over misinformation, election interference, and financial fraud as deepfake technology advances.

3. Hugging Face researchers have introduced Open Deep Research, an open alternative to OpenAI’s deep research tool, leveraging OpenAI’s o1 model and an open-source agentic framework for web browsing, data analysis, and autonomous research.

4. Snap Inc. launches new AI text-to-image model, optimized for mobile devices, generates high-resolution images in just 1.4 seconds while reducing costs by running entirely on-device.

5. OpenAI is updating its o3-mini model to enhance transparency by revealing more reasoning steps in ChatGPT responses, addressing competition from DeepSeek and improving clarity for free and paid users.

6. OpenAI has introduced European data residency, enabling organizations to process and store AI data within Europe to comply with GDPR and local regulations, offering zero data retention for API users and in-region storage for ChatGPT Enterprise and Edu users.

7. The EU has issued guidance clarifying AI system definitions under its risk-based AI Act, requiring developers to assess compliance or face fines of up to 7% of global annual turnover; the non-binding 13-page document will evolve with AI advancements.

8. 谷歌 has released its 2024 Responsible AI Progress Report, detailing advancements in AI governance, risk management, and safety protocols.?

9. 亚马逊 will unveil its next-generation AI-powered Alexa on February 26, featuring enhanced conversational abilities, task automation, and a potential $5–$10 monthly subscription model.

10. OpenAI is deploying ChatGPT to 460K students and 63K faculty across 23 California State University campuses, its largest educational rollout, providing AI-powered tutoring, study guides, and administrative support.

11. Alphabet Inc. (Google’s parent co.) has removed its prior restriction on using AI for weaponry and surveillance, sparking backlash from human rights groups like Human Rights Watch.

12. OpenAI is evaluating 16 U.S. states for new AI data center campuses as part of Stargate, a $100Bn joint venture with Oracle and SoftBank aimed at bolstering U.S. AI infrastructure.

13. Amazon plans to spend over $100Bn in 2025, primarily on AI infrastructure for its AWS cloud division, marking a sharp increase from $78 billion in 2024.

14. Boston Dynamics is teaming up with the Robotics & AI Institute RAI Institute , to enhance reinforcement learning for its Atlas humanoid robot, aiming to accelerate skill acquisition through trial-and-error-based AI.

15. Figure is ending its partnership with OpenAI to develop proprietary AI for its humanoid robots, with CEO Brett Adcock emphasizing the need for vertically integrated AI over outsourced intelligence models.

?? Top Funding News:

1. The EU is investing $56M in OpenEuroLLM, an open-source multilingual AI model developed by a consortium of 20 European institutions and trained on EU supercomputers to support businesses, governments, and key industries while ensuring transparency and regulatory compliance.

2. TrueFoundry , an AI infrastructure company automating AI deployment, raised a $19M Series A led by Intel Capital , with participation from Peak XV Partners , Eniac Ventures , Jump Capital , and angel investors.

3. Pendulum Systems , an AI-powered supply chain optimization platform, raised $22M from Lowercarbon Capital , Cross-Border Impact Ventures , Decisive Point , Collab Fund, and Blaise Aguera y Arcas to enhance its self-correcting AI for inventory, procurement, and pricing decisions.

4. &AI , developer of "Andy," an AI agent for patent attorneys that dynamically adapts to context to assist with tasks like drafting claim charts and office action responses, raised a $6.5M seed round led by First Round Capital , with participation from Y Combinator , BoxGroup , SV Angel , and angel investors.

5. DevAI , which develops agentic AI solutions like the Network Intelligence Engine and Neo to assist enterprise IT teams, raised a $6M seed round led by Emergence Capital , with participation from Pear VC , Base10 Partners , and Benchstrength .