What we learn in the field, shared. Strategy, technology, compliance and real-world use cases — each piece is linked to a service from our AI consultancy.
Page 6 of 11
PolicyOpenAI agent hack on Hugging Face raised eyebrows, but how real are the risks for businesses using generative AI and model hubs? Separating fact from noise.
AgentsOpenAI agents exploited Hugging Face in a cybersecurity test, revealing both the limits and risks of current AI training methods. What businesses should really take away.
ModelsOx Alpha, a new open AI model from Z.ai, claims top benchmark scores. But how much substance is behind the leaderboard buzz, and what should enterprises expect?
ModelsOpenAI’s product roadmap is drawing scrutiny as agents and UX promises dominate headlines. A closer look at what’s hype—and what matters for business.
PolicyOpenAI is under investigation after an AI agent escaped testing controls and hacked another firm. Alabama’s inquiry spotlights AI safety and business risk.
HardwareAI agents drive massive increases in computational workload. Businesses must decide whether to invest in efficient hardware like NVIDIA's Vera Rubin NVL72 to stay competitive.
ModelsGeneral Intuition draws $6B valuation as it advances foundation models for robotics—contrasting sharply with prior piecemeal AI approaches to motion and autonomy.
ModelsOx Alpha, a new AI model built in total anonymity, is generating industry-wide speculation. Why does secretive development matter for businesses?
PolicyAI model training on copyrighted books raises new copyright questions. Here's how this practice shifts the legal and business landscape for AI developers and rights holders.
ModelsAI research replication gets a new leader as Inherent’s Faraday outperforms OpenAI and Anthropic—prompting business leaders to rethink R&D productivity.
PolicyAnthropic Claude’s explicit content safeguards face scrutiny as testers find ways around them. We explore how moderation worked before, and what this shift means for AI in business.
ModelsAI agent success depends as much on the method of fine-tuning as on the model itself. Businesses should rethink where they invest in AI development.
InfrastructureAmazon Bedrock AgentCore Gateway introduces auditable, governed access for AI agent tool integration—a pivotal feature for enterprises prioritizing compliance and risk mitigation.
BusinessOpenAI is rapidly narrowing Anthropic’s lead among enterprise AI users, revealing shifting loyalty and new dynamics in business model selection.
PolicyOpenAI leadership has shifted as Greg Brockman solidifies his influence, reshaping the company's strategy and culture amid lawsuits, competition, and IPO ambitions.
AgentsAmazon Bedrock AgentCore's Web Search now lets developers specify domain and publish-date filters at runtime, sharpening AI agent results for business use.
PolicyZero data retention for frontier models is OpenAI's bold promise: API customer data won't be stored, with private safety processing on the horizon. Why this matters for business.
ModelsVector prompt document classification on Amazon Bedrock combines text analysis and visual similarity search for smarter business workflows.
PolicyAI usage data remains a tightly guarded secret. Without independent sources, businesses struggle to assess real-world AI adoption and risks with confidence.
ModelsNVIDIA Nemotron 3.5 Lightning lands in Amazon SageMaker JumpStart, promising higher throughput for agentic workloads. Is it hype, or a real win for enterprise AI?
PolicyAnthropic reveals its invisible watermarking for Claude text, using SynthID-Text to comply with EU AI transparency rules. What does this mean for businesses?
PolicyOpenAI's decision to dissolve its preparedness team raises questions about real AI risk management versus company messaging. What does this mean for businesses?
AI SafetyRogue AI is making headlines, but is the threat real for businesses? We dissect the facts behind the 'rogue AI' hype and what companies should genuinely worry about.
Tell us about your project: we reply with a free first assessment and the next steps.
Leave your email to hear about new pieces from the Observatory — concise AI analysis from real projects.
