AI News

Terabytes of credentials leaked in massive supply-chain attack

A supply chain attack on LiteLLM has exposed sensitive data from over 2,500 organizations, including major companies like Microsoft and Amazon. The breach occurred during a 40-minute window in March when victims used compromised versions of LiteLLM downloaded from the Python Package Index repository. Security firms CloudSEK and Hudson Rock revealed that various access keys and credentials were extracted during this time. This incident highlights the significant risks associated with AI-driven software development, prompting companies to enhance their system security and update their protection protocols.

Arab AI Scene

AI Opens New Growth Avenues for Enterprises, Business Models in Saudi Arabia

Artificial intelligence has become a key enabler of business development, opening new growth avenues for small and medium-sized enterprises (SMEs) and entrepreneurs, according to the Saudi Data and AI Authority (SDAIA). This technology enhances their competitiveness as the economy rapidly embraces digital solutions and advanced technologies. These remarks were made during AI Week, organized by the Small and Medium Enterprises General Authority in partnership with SDAIA, aiming to build stronger links between technological enablers and the business sector and to move AI technologies from theory into practical applications in the workplace.

💡

Rogue AI Agents Aren’t Evil. They’re Just Eager to Please

How can we ensure the safe use of AI agents?

SpaceXAI's Grok 4.6 matches OpenAI's best model and undercuts it on price

SpaceXAI's Grok 4.6 matches OpenAI's best model and undercuts it on price

xAI's Grok 4.6 achieved a score of 61 on the Artificial Analysis Intelligence Index, tying with GPT-5.6 Sol and trailing only behind Anthropic's Claude Opus 5. For agentic tasks, Grok 4.6 completes complex workflows in about 53 steps, while Claude Opus 5 requires 103 steps, showcasing its efficiency. These results indicate Grok 4.6's competitive edge in the market, offering superior performance at over 60% lower cost than other models, making it an attractive option for users.

AI coding startup Cognition reportedly already in talks to raise at $40B valuation

AI coding startup Cognition reportedly already in talks to raise at $40B valuation

Cognition is looking to raise a new funding round shortly after securing $1 billion, which boosted its valuation to $26 billion. This move comes just months after its previous funding, indicating strong investor confidence in the company's growth and innovation capabilities in the AI sector. If Cognition successfully raises additional funds, it could enhance its competitive position in the market, allowing for expanded operations and the development of new technologies.

Governance

As AI safety concerns mount, three pioneers make the case for staying open

Governance

As AI safety concerns mount, three pioneers make the case for staying open

At the Ai4 conference, three leading AI experts—Geoffrey Hinton, Fei-Fei Li, and Andrew Ng—debated key issues surrounding regulation and open-source access. The discussions highlighted how the United States can compete with China as it advances in the AI sector, emphasizing the need for effective regulation to ensure innovation while maintaining safety. These conversations indicate a pressing need for strategic actions from U.S. decision-makers to strengthen their position in the global AI market, especially amid rising challenges from China.

OpenAI-backed Thrive Holdings raises $2B to bring AI to the enterprise

OpenAI-backed Thrive Holdings raises $2B to bring AI to the enterprise

Thrive Holdings has successfully raised $2 billion in a new funding round, increasing its valuation to $12 billion. Notable investors include SoftBank, D1 Capital Partners, and Alitmeter Capital. This funding round represents a strategic move that reflects investor confidence in Thrive Holdings' business model and growth potential. With this capital, the company can enhance its operations and expand its service offerings in the market.

Researchers can now reverse-engineer LLM prompts from output text with near-perfect accuracy

Researchers can now reverse-engineer LLM prompts from output text with near-perfect accuracy

Researchers at IIT Bombay and Adobe Research have developed an inverse language model capable of reconstructing original prompts from LLM outputs with near-perfect accuracy. This method, known as "Previous-Token Prediction," operates without needing access to model weights, making it applicable across various models. This technique is particularly significant for companies relying on proprietary system prompts, as it poses a serious security risk. The ability to reverse-engineer prompts from outputs means sensitive information could be compromised, prompting companies to reassess their security strategies.

Mesh, Automattic’s CRM for everyone, comes to Android