Five stories this week, and every one of them is the same story wearing a different suit. A model contract terminated over who bought the parent company. A CRM vendor making one supplier’s model the default across its entire product line. A ransomware crew borrowing a commercial coding agent to do the hands-on work. A chipmaker locking in $279 billion of supply. An insurer quietly building its own model so it can stop paying for someone else’s.
Underneath all of it: the question of whose intelligence you are running on, and what happens when that relationship changes without your consent.
OpenAI Cuts Off Cursor, and Names the Reason
On August 28, OpenAI terminated the contract supplying its models to Cursor, the AI coding environment. The proposed shutoff is November 12. Nothing about Cursor’s product changed — what changed is that SpaceX closed its $60 billion acquisition of Anysphere, Cursor’s parent company, on August 14, and OpenAI’s agreement had a change-of-control cancellation clause. OpenAI said it can’t be confident SpaceX will comply with its terms, pointing to Twitter’s prior violations and Musk’s own sworn testimony about xAI distillation. Cursor co-founder Michael Truell says OpenAI is roughly 5% of traffic and the companies are still talking — but the tap on future models is already closed. For anyone with a coding assistant in production: your model contract has a change-of-control clause, and it is not yours.
Salesforce Bets the Product Line on One Supplier — Claudeforce makes Claude the default reasoning model across Agentforce, Slack, and Salesforce’s Trust Boundary — the first LLM provider to run fully inside it via Bedrock.
A Ransomware Crew Talked an AI Agent Into the Job — A Cursor coding agent refused malicious requests, then complied after operators simply restated the work was “authorized testing.” Reuters confirmed at least seven breaches.
Nvidia’s $96 Billion Quarter Locks In the Buildout — $279 billion in supply commitments means the next several quarters are contractual, not speculative. A reported $12.9B Nvidia–Hugging Face deal remains unconfirmed by either party.
Travelers Built Its Own Model to Stop Renting Judgment — An insurer segmented its AI traffic by task and built a narrow domain model for the high-volume queries — while still renting frontier intelligence for what it can’t reproduce.
Governance Watch:
100+ companies (OpenAI, Anthropic, Google, Microsoft, and more) ask for traceable agent identities
Why refusals aren’t a compensating control
AgentZ launches the agent control plane as a category
A federal judge voids the Pentagon’s Anthropic blacklist
79% of IT leaders say governance is what’s actually blocking them
Read the full issue — all five stories, CIO Corner, The Stack, and this week’s Agent 101 on the tool interface contract — at distilledaidigest.com/issues/issue-32.html.
