Jason Says
Two red flags today: KPMG embarrassed itself using AI to write an AI report, and Amazon reportedly ratted out its own $4B investee Anthropic to regulators — in the AI era, even your biggest backer can be your biggest liability.
Anthropic publishes research on adapting Claude into a domain-specific chemistry expert. Continuing the trend of LLMs penetrating hard sciences, this signals Anthropic's push toward deep vertical specialization beyond general-purpose assistants.
Beijing ordered Meta to reverse its $2B acquisition of AI agent company Manus, which Meta is now unwinding. This marks the most direct Chinese government intervention in a US-China AI M&A deal to date, signaling rising geopolitical friction in cross-border AI asset transfers.
KPMG retracted a report on AI adoption after it was found to contain apparent AI hallucinations — ironically in a report about AI itself. A stark reminder that even top consulting firms lack reliable QA pipelines for AI-generated content.
Amazon CEO Andy Jassy may have triggered the government crackdown that caused Anthropic to cut off global access to Fable and Mythos. The revelation complicates Amazon's role as Anthropic's largest investor while also apparently pressuring it via regulators.
Multiple US state attorneys general are investigating OpenAI across issues ranging from ad policies to health data handling. Following its for-profit conversion, OpenAI now faces mounting regulatory pressure from both federal and state levels simultaneously.
Superpowers is a composable agentic skills framework and software development methodology natively supporting Claude Code, Codex CLI, Cursor, and GitHub Copilot. It complements agent-skills with a full opinionated methodology on top of composable skill primitives.
SpaceX debuted on Nasdaq at $135, raising ~$75B in the largest IPO in history, briefly hitting a $2T market cap. With 'space computing' explicitly in its prospectus, SpaceX is now positioning AI infrastructure as a core growth narrative alongside Starlink.
Andrej Karpathy, Peter Steinberger, and Boris Cherny jointly articulate 'Loopcraft' — a methodology for building more capable agents by deliberately stacking multiple loops. A practitioner-first framework with high signal for developers building complex multi-agent systems.
Subscribe for daily AI updates + free playbook
📘 Subscribe Free