AI News

OpenAI CTO Mira Murati revealed that the Sora text-to-video model could debut soon. She confirmed the use of publicly available and licensed data but was vague on specifics regarding platforms like YouTube and Facebook, raising concerns amid ongoing copyright lawsuits. This lack of clarity may affect trust, as creators question the use of their work in AI training, highlighting the need for transparency in generative AI practices.

Invoke has launched Invoke Workflows, a suite of AI tools for game developers, simplifying AI adoption for non-technical team members. With features for training assets and secure image generation, it caters to large studios, emphasizing an artist-centric approach to enhance creative workflows.

Databricks partners with Paris startup Mistral, investing undisclosed funds in its Series A to integrate Mistral's open-source LLMs into its data platform. Users can access advanced generative AI models while maintaining security standards. Mistral, also partnering with major companies, is gaining traction in the AI space.

The European Parliament has approved the AI Act to regulate artificial intelligence, aiming to protect rights and foster innovation. With broad support, it will enforce strict rules on high-risk AI applications, including bans on certain uses like biometric categorization. Violations may incur severe fines. The act awaits final checks before implementation.

Businesses need to ensure AI models are safe for deployment, emphasizing thorough testing to avoid mishaps. Kolena's AI Quality Platform streamlines testing, enabling developers to rigorously validate models against real-world scenarios, addressing bias and performance issues. The platform is being refined via beta testing with Fortune 500 companies.

San Francisco startup Anthropic has launched Claude 3 Haiku, its fastest and most cost-effective AI model with advanced vision capabilities. Designed for high-volume applications, it processes 21,000 tokens per second and enhances enterprise efficiency with a focus on security, versatility, and performance across various sectors.

Proscia has raised $9 million in a Series C extension, bringing total funding to $46 million, to bolster its digital pathology solutions after FDA clearance. This supports the company's growth in the U.S. market, enhancing AI applications for diagnostics amid rising cancer rates and a pathologist shortage.

Microsoft's Copilot for Security, launching globally on April 1, uses generative AI to assist cybersecurity professionals amidst a talent shortage and rising cyber threats. It offers a centralized dashboard with AI insights and is built on OpenAI’s GPT-4 model, aiming to enhance human capabilities and streamline security efforts.

Figure, a robotics startup valued at $2.6 billion, revealed its humanoid robot, Figure 01, in collaboration with OpenAI. The robot can interact, retrieve items, and perform tasks. CEO Brett Adcock highlighted the need for further development before commercial release, aiming for a billion-unit scale while avoiding harmful uses.

The AI Impact Tour in Boston on March 27 will explore generative AI applications in enterprises, highlighting data integrity. Speakers include Caroline Arnold of State Street and David Clifford of Biogen, discussing challenges and solutions in finance and biotech. Limited to 50 attendees, the event will provide networking opportunities and insights into building reliable AI models.

Lawmakers in the European Parliament have approved the AI Act, the first comprehensive regulation for high-risk AI systems. Set to take effect in May, it requires global compliance, impacting U.S. companies. Experts highlight its role as a standard for trustworthy AI and call for immediate compliance efforts.

Cerebras and G42 have launched the Condor Galaxy 3, an AI supercomputer with eight exaFLOPs performance, powered by 58 million AI-optimized cores. This model features the WSE-3 chip, enabling efficient training of large AI models and aims to exceed 55 exaFLOPs by 2024.

NASA has made a historic move by appointing its first chief AI officer, David Salvagnini. This new role expands his responsibilities as the space agency's chief data officer. Salvagnini is charged with guiding NASA's initiatives to ensure the responsible deployment of artificial intelligence (AI), focusing on the development of risk management tools and essential training programs.

The Ray framework, crucial for scaling AI workloads, has a critical security flaw dubbed "ShadowRay," allowing attackers to access sensitive data since it lacks formal patching. This vulnerability raises concerns for organizations using Ray, including major companies like Amazon and OpenAI, highlighting the need for better security practices.

Google's AI chatbot, Gemini, temporarily halted its image generation feature following backlash over historically inaccurate portrayals, like depicting U.S. Founding Fathers as Black. The company aims to refine the model for better accuracy while addressing user concerns about diversity and bias.