Price:

AI & TECH

Industry rewrites Trump AI order to speed releases, curb China

Sunday, June 7, 2026 · from 1 podcast, 2 episodes
  • David Sacks pushed Trump to rewrite an AI order, slashing government review from 90 to 30 days.
  • OpenAI’s Codex now grows fastest among non-technical workers, shifting focus to orchestrating parallel workflows.
  • Microsoft pivots from pure performance to enterprise cost-cutting, betting on 10x cheaper specialized models.

The new AI executive order, finalized last week, is less a government mandate than a formalized handshake. After a last-minute call from investor David Sacks, President Trump scrapped a draft requiring 90-day government reviews of advanced AI models. The final version cuts that to 30 days and explicitly forbids a mandatory licensing regime.

The administration treated the regulation like “eating its vegetables,” says host Nathaniel Whittemore. It creates a paper trail for safety hawks like Steve Bannon while protecting industry interests by keeping participation voluntary. The order tasks the NSA with primary testing of models with meaningful cyber capability jumps, like Anthropic's Mythos, but lacks any mechanism to block a release. According to Whittemore, this effectively codifies deals already in place between the government and major labs.

“The policy process looked more like a chaotic negotiation than a standard signing ceremony.”

- Nathaniel Whittemore, The AI Daily Brief

Anthropic is walking back its own aggressive timelines. Despite promising a Mythos-level model in weeks, the company now says robust safeguards to prevent cyber misuse “do not yet exist.” It has, however, expanded access to 150 new partners across critical infrastructure sectors like energy and healthcare under Project Glasswing.

OpenAI’s growth reveals where the real enterprise action is. Codex now has 5 million weekly active users, with non-technical knowledge workers adopting it three times faster than developers. The company is redesigning the tool to manage the “strange abundance” of modern work - finding information across systems and coordinating approvals. A new “Sites” feature lets users turn any AI-generated artifact, like a revenue forecast, into a disposable, shareable web app.

“Users can build an event dashboard or a revenue forecast planner that lives as a URL rather than a file.”

- Nathaniel Whittemore, The AI Daily Brief

Microsoft is betting the next enterprise battle isn’t about raw intelligence but cost. Its new MAI Thinking-1 model lags behind frontier models on some benchmarks but, when tuned for specific tasks, claims to outperform GPT-4.5 at one-tenth the cost. CEO Satya Nadella advocates for enterprises to move from consuming frontier models to building cost-optimized proprietary ones. Companies like Uber are already capping employee token spending at $1,500 monthly.

The order and the product shifts point to a single reality: the industry is prioritizing speed and affordability over restrictive oversight, with an eye firmly on beating China.

Source Intelligence

- Deep dive into what was said in the episodes

What OpenAI and Anthropic Think Happens Next With AIJun 5

  • Nathaniel Whittemore says Trump's AI executive order evolved from mandating 90-day pre-release government access to a voluntary 30-day process, with NSA assigned primary testing responsibility.
  • David Saxs intervened to stop the initial executive order, arguing it would hinder US AI competitiveness against China. The final version includes a disclaimer prohibiting mandatory government licensing or pre-clearance regimes.
  • Dean Ball calls the executive order a major win for AI safety advocates within the administration and a significant loss for accelerationists like Saxs, arguing it tees up infrastructure for future mandatory licensing.
  • Anthropic expanded Project Glasswing access to its Mythos model, adding 150 partners across 15 countries including energy, water, communications, healthcare, and hardware sectors vulnerable to catastrophic cyber attacks.
  • Anthropic walked back its timeline for general Mythos access, stating robust safeguards preventing cyber capability misuse don't exist yet. Current testers find the model powerful but prohibitively expensive, with Anthropic subsidizing costs.
  • OpenAI reports Codex now has 5 million weekly active users, with non-technical knowledge workers adopting it three times faster than developers. The platform sees users shifting from sequential to parallel task execution.
  • OpenAI identifies three knowledge work frictions: finding inputs across opaque systems, information coordination costs, and approval delays. A McKinsey study found workers spend over 25% of their week on email and nearly 20% searching for internal information.
  • OpenAI's new Codex features include annotations for precise document interaction, role-specific plugins bundling apps and skills for six functions, and Sites for turning artifacts into shareable web apps.
  • Uber implemented a $1,500 monthly token spending cap per employee, highlighting cost management as a critical vector in enterprise AI adoption amid the broader shift from subsidy to scarcity economics.
  • Microsoft released seven new AI models including MAI Thinking One, a 1-trillion parameter MoE model positioned between Sonnet 4.6 and Opus 4.6. Mustafa Suleyman claims it outperformed GPT-4.5 on quality while being 10x lower cost for McKinsey tasks.
  • Microsoft's strategy focuses on Frontier Tuning for company-specific agents, with CEO Satya Nadella advocating for enterprises to move from consuming frontier models to participating in the frontier ecosystem via cost-optimized proprietary models.

The Next Wave of Enterprise AIJun 3

  • President Trump's recent AI executive order requires companies to voluntarily share advanced models with the government 30 days before public release, a reduction from the draft's 90-day period. The NSA is tasked with primary testing.
  • David Sachs intervened to stop the initial executive order signing. He later clarified the policy only targets models with a meaningful step change in cyber capabilities, like Anthropic's Mythos, not incremental updates.
  • The executive order expressly prohibits creating a mandatory government licensing or preclearance regime for AI models, a direct response to critiques of a de facto licensing system.
  • Anthropic expanded Project Glasswing, granting 150 new partners across 15 countries access to its Mythos model. The rollout now includes energy, water, communications, healthcare, and computer hardware sectors.
  • Anthropic stated robust safeguards to prevent misuse of Mythos's cyber capabilities do not yet exist, walking back earlier promises of a near-term public release. Testers find the model powerful but extremely expensive.
  • OpenAI's Codex hit 5 million weekly active users, with non-technical knowledge workers adopting it three times faster than developers. 72% of its knowledge worker users produce an artifact weekly.
  • OpenAI's report identifies three core frictions in knowledge work: finding inputs across systems, information coordination costs, and approvals. It argues Codex acts as a factory redesign to address these.
  • About 50% of Codex users now run multiple tasks in parallel during the day, a shift from sequential use that lets a single worker operate at the scale of a small team by orchestrating workstreams.
  • OpenAI's Codex update introduced annotations for precise document interaction, role-specific plugins bundling apps and skills for six functions, and a Sites feature to turn artifacts into shareable web apps.