OSASI.ORGCommentary · Ideas for our profession
All briefs← Back to OSASI

Cheaper isn't private: the labs solved cost, not sovereignty

October 13, 2026 · OSASI Commentary

The labs have solved cost, The AI Daily Brief says, but not sovereignty. Two answers are on offer: OpenAI's Private Intelligence, and open-source agents like Hermes.

How a brief is made

A member reads an article or listens to a podcast and sends the snippets, key thoughts or concerns tied to a passage or a timestamp. An AI assistant, chosen for the task and the setup (Claude, ChatGPT, Codex, Gemini or another), searches the full article or transcript, pins the times, checks the quotations, and assembles a draft. The draft goes through rounds of review, and nothing is published until a person approves that exact version. The cues on each post start the player at the passage discussed. Commentary is OSASI's perspective, offered for discussion, not a summary of the source. Each brief is co-authored by a person and an AI assistant. AI assistant for this brief: Claude (Anthropic).

Source podcast · The AI Daily Brief

This episode prompted the commentary below. It is one perspective, offered for discussion.

The AI Daily Brief · October 8, 2026
The Most Important Trends in New AI Products

Listening highlights · Nathaniel Whittemore, host, on his own. Times from Spotify's transcript and chapters.

−25:05 to −24:02 remaining"You should be able to"Nous Research's list: local models, your choice of data, a machine that never touches the internet
−4:21 to −3:24 remainingCost versus sovereignty"If you want the cheapest LLMs, you should now buy American", and what that leaves unanswered

Tap a time above to start the player there. If Spotify shows a card about the show first, close it and the player carries on at that time. Times are approximate transcript times; ads and episode editions may shift them. This plays the full episode, not an extracted clip.

Cost may be solved

Claude Haiku 5.5 runs at a quarter of Sonnet 5.5's price, about 21 cents per task in Artificial Analysis' tests. Whittemore's verdict is that the closed frontier labs "have increasingly solved the cost side of that equation", but not sovereignty. OSASI would put that second half more cautiously: it remains to be determined, as users balance cost with privacy.

OpenAI's answer: Private Intelligence

At DevDay, OpenAI added Private Safety Processing to what it calls Private Intelligence (The Artificial Intelligence Show, 7:20). Why is that new, if OpenAI already promised privacy? Because the promise came in tiers. By default, the API keeps abuse-monitoring logs for up to 30 days. Zero Data Retention, where nothing is kept, needs OpenAI's approval, and OpenAI can withdraw it for a model to investigate "severe risk activity", then keep and human-review flagged content.

OpenAI says frontier models made that harder, because some misuse only shows across many requests. Private Safety Processing is its fix: flagged records stay encrypted in the customer's storage for 30 days, and automated checks send OpenAI a signal, not the content. It is for approved enterprise API accounts, still rolling out, with no independent review published yet. It answers who at the lab can see your data, not whose model, servers or terms.

The open answer: Hermes Agent

Nous Research's list of what you "should be able to" do with an agent includes running it "on a machine that never touches the internet" and knowing "what it did and when and why." Most of the list is more setup than an appraisal office needs. Those two items are what the job already asks for.

Send a thought →

A like is counted once per reader per day and stores nothing about you. A thought goes by email to the person who edits these briefs; nothing you write appears on the site unless it becomes part of an approved brief. Share opens the service you pick; this site adds no trackers.

Prefer a feed reader, Slack or Teams? Follow by RSS.