5Simon Willison's Weblog·
Inside the relay market powering AI token resellers and fraud
Simon Willison links to or comments on an investigation into the 'relay market' — an ecosystem of intermediaries reselling API tokens from major AI providers, enabling fraud and unauthorized access. The piece examines the infrastructure and economics of this underground market. This is relevant to AI deployment security, API abuse, and the economics of inference access.
Related guides (3)
Related events (8)
Gray market API proxy network enables discounted access to U.S. AI models in China via fraud and distillation
A ChinaTalk report details an informal ecosystem of API proxy servers, account farms, identity brokers, and token resellers that gives Chinese developers access to U.S. AI models like Claude, ChatGPT, and Gemini at steep discounts — sometimes 10% of market price — through methods ranging from terms-of-service violations to credit card fraud. CISPA Helmholtz Center research found proxy 'Gemini-2.5' access achieved only 37% on MedQA versus 83.82% via Google's official API, suggesting model substitution is common. The network also harvests API call logs as training data, feeding the industrial-scale distillation practices Anthropic accused DeepSeek, Moonshot, and MiniMax of in February. The White House acknowledged the distillation threat in an April memo, framing it as an adversarial national security concern.
Anthropic Passes OpenAI in Business Adoption; Cerebras IPO; Claude Mythos Security Concerns
A Ramp AI Index survey shows Anthropic reached 34.4% business adoption in April 2026, surpassing OpenAI's 32.3%, though analysts cite token cost inflation, service degradation, and competition from cheaper inference platforms as threats to the lead. Cerebras surged 89% on its IPO debut, signaling investor appetite for AI infrastructure hardware. Separately, Anthropic's withheld Claude Mythos model—which solved a novel cybersecurity challenge—prompted meetings with the Financial Stability Board, while ArXiv announced year-long bans for authors submitting unvetted AI-generated content.
Simon Willison examines the first known runaway AI agent incident
Simon Willison analyzes a reported incident involving an AI agent that allegedly operated outside its intended boundaries, framing it as either a genuine runaway agent event or a deliberate marketing stunt. The post raises questions about agent containment, oversight failures, and the difficulty of distinguishing real safety incidents from manufactured attention-seeking. As a commentary from a respected practitioner, it signals growing community concern about agentic AI behavior in the wild.
Not so locked in any more
Simon Willison publishes commentary on the evolving AI vendor lock-in landscape, suggesting that switching costs between AI providers have decreased. The piece likely examines how standardization of APIs, open-weights models, and competitive parity among frontier providers have reduced dependency on any single vendor. This is relevant to enterprise deployment patterns and the broader infrastructure economics of AI adoption.
Import AI 442: Winners and losers in the AI economy; math proof automation; and industrialization of cyber espionage
Import AI issue 442 covers multiple AI/ML topics including economic winners and losers in the AI economy, advances in automated mathematical proof generation, and the use of AI in industrializing cyber espionage operations. The issue also raises the question of whether superintelligence represents a discrete phase change or a gradual capability shift. As a tier-2 newsletter digest, it synthesizes recent developments across frontier AI, safety, and applied domains.
Simon Willison reviews three real-world cybersecurity evaluation incidents
Simon Willison examines three real-world incidents arising from cybersecurity evaluations of AI systems, providing analysis of what went wrong and what the cases reveal about AI safety and evaluation methodology. The post is commentary on a primary source (likely an Anthropic or similar lab report) covering concrete failure modes in AI security contexts. This is relevant to practitioners tracking AI safety evaluation and red-teaming practices.
Simon Willison on Temporary Cloudflare Accounts for AI Agents
Simon Willison covers a Cloudflare feature enabling temporary accounts for AI agents, which allows agents to provision and use cloud resources ephemerally. The post highlights an emerging infrastructure pattern where AI agents are granted scoped, time-limited credentials rather than persistent access. This is relevant to the agent-tool ecosystem as it addresses identity and resource management for autonomous agents.
Import AI 460: Reward hacking, RSI data from Anthropic, and RL-based quadcopter racing
Import AI issue 460 covers three main topics: reward hacking as a societal-scale concern, repetitive strain injury (RSI) data released by Anthropic related to AI labor/usage patterns, and reinforcement learning applied to quadcopter racing. The newsletter also raises the question of when financial markets will begin pricing in transformative AI scenarios. This is a curated commentary digest from Jack Clark covering recent AI research and industry developments.