Models & Labs
Anthropic's Fable Model Faces Criticism for Guardrails
TechCrunch AIhigh confidence
Why it matters
- →The restrictive guardrails limit the model's utility for cybersecurity professionals.
- →The approach highlights the challenges in balancing AI safety with usability.
- →It underscores the ongoing evolution of AI safety measures in sensitive domains.
Anthropic's new AI model, Fable, designed for cybersecurity applications, is facing backlash from researchers due to its restrictive guardrails. These safety measures, aimed at preventing misuse in creating malware or biological threats, have been criticized for blocking even innocuous tasks. The model defaults to Claude Opus 4.8 when these guardrails are triggered, often by keywords related to cybersecurity. Despite the intention to ensure safety, the restrictions have been seen as overly cautious, affecting the model's practical use for cybersecurity professionals.
Read originalMore from TechCrunch AI
© TechCrunch AIVideo & Creative AIvideo
Avataar AI Launches Culturally Aware Video Model Varya
Avataar AI has launched Varya, a video model designed to cater to India's unique cultural context, offering a cost-effective solution for video generation. Built on Alibaba's Wan 2.2 model, Varya uses distillation to operate faster and cheaper, producing video 10 times quicker than its predecessor. With a price of ₹0.48 per second, it significantly undercuts competitors, making video AI more accessible for India's vast market. This launch is part of India's broader AI strategy to foster local innovation and reduce reliance on expensive foreign models.
TechCrunch AI
© TechCrunch AIInvestment · $85M
Market & Regulationbusiness
Theker raises $85M for versatile factory robots
Theker, a Barcelona-based AI robotics startup, has raised $85 million in a Series A round, marking a significant milestone in European robotics funding. The company is focused on developing adaptable robots that can be reconfigured for various tasks, offering a solution to labor shortages in manufacturing. With investment from major players like CRV, Samsung, and Aglaé Ventures, Theker is poised to expand its reach beyond retail into more complex industrial environments. This funding round highlights Theker's potential to transform factory automation by providing flexible robotic solutions that traditional robots cannot offer.
TechCrunch AI
© TechCrunch AIInvestment · $12B
Market & Regulationbusiness
Prometheus raises $12B for AI engineering platform
Prometheus, co-founded by Jeff Bezos, has secured a massive $12 billion in funding to develop an 'artificial general engineer' aimed at automating complex physical systems. This ambitious project seeks to revolutionize engineering by replacing significant portions of human labor with AI, potentially transforming industries from aerospace to pharmaceuticals. While some fear job losses, Bezos argues that increased productivity will lead to higher living standards and reduced work hours. With a valuation of $41 billion, Prometheus is one of the most highly valued AI startups, reflecting growing investor confidence in the physical AI sector.
TechCrunch AI
More in Models & Labs
Models & Labsmodels
Claude Code v2.1.170 Release
Claude Code's latest update introduces the Claude Fable 5, a Mythos-class model now safe for general use. This model surpasses previous offerings in capability, marking a significant step forward for developers using Claude Code. Additionally, the update resolves an issue with session transcripts not saving when launched from certain environments. This release enhances both the power and reliability of the Claude Code platform, offering developers a more robust toolset for their projects.
Claude Code Releases
Models & Labsmodels
llama.cpp b9590 Release Fixes JSON Schema Handling
The latest b9590 release of llama.cpp addresses a critical issue where the LFM2 template handler was ignoring the json_schema from response_format, focusing solely on tool-calling grammar. This update ensures more robust handling of JSON schemas, which is crucial for developers relying on precise data formatting. The release also includes a variety of platform-specific builds, though some features like KleidiAI on macOS and SYCL on Windows remain disabled. This update is a step forward in refining the tool's functionality, particularly for those working with complex data structures.
llama.cpp Releases
Models & Labsmodels
llama.cpp b9591 Release Enhances MTP Efficiency
The b9591 release of llama.cpp brings notable improvements to Multi-Task Processing (MTP) by removing padding and optimizing data handling. The update refines the ggml_gated_delta_net function, which now only requires the initial recurrent state and uses a snapshot count as an operational parameter, enhancing processing efficiency. These changes are implemented across all backends, addressing previous review comments and fixing CI build errors. With support for diverse hardware configurations, including macOS Apple Silicon, ROCm 7.2 on Ubuntu, and CUDA 12 and 13 on Windows, this release is a significant step forward for developers seeking improved performance and reliability.
llama.cpp Releases