Paul Goldman

Founder & CEO @itmethods | Governed AI. Outcomes assured. | Operate · Build · Govern · Assure | Banking · capital markets · high-tech · defence · life sciences

[Paul Goldman @paulgoldman1]

Safety done properly is not a gate. It is a throughput system.

Agree the boundary in advance, tier it by consequence, and a new AI use case lands in a tier instead of going to committee from scratch.

The rails are what let you go fast.

https://t.co/A9Q8mgg9Hv

[Paul Goldman @paulgoldman1]

The Answer Is an Estimate. Somebody Has to Check It. https://t.co/9iXsTZ81kH

[Paul Goldman @paulgoldman1]

The filing is not the work.

CRA Article 14 starts 11 September. 24-hour early warning. The industry will treat it as a filing problem. That work is real. It is not the exam.

A shared fix is not proof it landed. Can you produce the artifact?

https://t.co/szPCFsxIbT

[Paul Goldman @paulgoldman1]

CEO Alex Karp of @PalantirTech is right that you have to own the application layer. The unresolved part is what owning it means when the layer is somebody's product.

Off a frontier lab and onto a vendor is a change of counterparty, not a change of posture.

Full piece in The Trust Layer: https://t.co/CMa8Cf2vKF

[iTmethods @itmethods]

Your agents are shipping code. Who is proving it was allowed?

The Dark Factory: agents build it, a human approves every merge, Reign proves it. Regulator-ready by default.

15s below → https://t.co/2GVZcTXNMZ

[Paul Goldman @paulgoldman1]

Karp on @CNBC yesterday: trust returns when AI customers can ask five questions about data, caching, prompts, weights, and alpha.

Microsoft today: a 2.5 billion dollar learning loop in which human capital and token capital compound.

The loop is real. The question is whose tenancy it compounds in, and what leaves with you at exit.

Full framework next week in The Trust Layer. https://t.co/v5XxQa4zZl

[Paul Goldman @paulgoldman1]

@Muskoka411 fogged up!

[Elon Musk @elonmusk]

Happy Father’s Day!

[Paul Goldman @paulgoldman1]

Canada is finally building sovereign AI. One layer is still to build. Every named champion is infrastructure, models, or research. The control layer above the model has no Canadian champion.
Last Friday proved that is the layer that decides everything.

https://t.co/48Z6wMiX3e

[Paul Goldman @paulgoldman1]

On Friday at 5:21pm ET, the most capable AI model on the public market disappeared. Not throttled. Gone, pulled off the market by a U.S. export-control order, not a vendor call.

Sovereignty isn't owning the best model. It's never being hostage to one.

New piece: https://t.co/XJ27XB0huJ

[David Sacks @DavidSacks]

I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true:

— As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable.

— Fable is Mythos with guardrails. But if those guardrails fail, then you’ve exposed Mythos and its advanced cyber capabilities to people who shouldn’t have them. (Keep in mind that Anthropic itself widely promoted the idea that Mythos was a cyberweapon and needed to be regulated as such. They asked for government regulation of Mythos and championed the guardrails on Fable. If there is a vulnerability — big or small — it is Anthropic’s responsibility to patch.)

— A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails. The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused.

— In their blog post, Anthropic defended its decision by saying the jailbreak isn’t serious. That is not what the trusted partner and the USG believe; nor is that kind of minimizing language consistent with Anthropic’s brand as the AI safety company. It’s difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not “serious.”

— In the past, Anthropic has always said that safety must be top priority and taken super seriously. In this case, Anthropic prioritized the continued offering of the consumer model over safety.

— In reaction, the Admin issued the export control. The Admin did this reluctantly. It’s been very surprised that Anthropic hasn’t wanted to cooperate with a reasonable safety request (ie fixing the jailbreak issue). Anthropic’s reaction is very much at odds with their branding and ethos as a safe AI research community.

— The Admin’s hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority.

— Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong. The Admin values Anthropic’s technical capabilities and feels that this issue, while serious, should be easily resolved. The ball is in Anthropic’s court.

[Cursor @cursor_ai]

We’re introducing the Cursor SDK so you can build agents with the same runtime, harness, and models that power Cursor.

Run agents from CI/CD pipelines, create automations for end-to-end workflows, or embed agents directly inside your products.

[Paul Goldman @paulgoldman1]

Posted this on the company account — the 114-day countdown is the part people keep missing

[iTmethods @itmethods]

1/5

114 days until EU AI Act high-risk enforcement.

Most enterprises are still in "policy and spreadsheet" mode.

Here's what regulators will actually ask for on August 2, 2026 🧵

[Elon Musk @elonmusk]

First (supervised) FSD approval in Europe!

Congratulations to the Tesla team and thank you to the regulatory authorities in the Netherlands for all the hard work required to make this happen.

[Paul Goldman @paulgoldman1]

Check out my latest article: Self-Hosted AI Agents Are Here. The Governance Isn't. https://t.co/piNcjuZgAU via @LinkedIn

[iTmethods @itmethods]

Check out our latest article: Securing the Agentic Era: OpenClaw, NemoClaw, and the Governance Layer That's Still Missing
#OpenClaw #NemoClaw #AIGovernance #Reign #EnterpriseAI #CISO #AgenticAI #GTC2026 #NVIDIA
https://t.co/OdMTfzuN5z

[SpaceX @SpaceX]

Announcing TERAFAB: the next step towards becoming a galactic civilization https://t.co/xTA70LOU0e

[Josh Kale @JoshKale]

Tesla’s TeraFab officially begins TODAY 🏭 This is make or break for possibly the whole AI industry

Elon’s trying to build a 2nm chip fab from scratch. The most advanced process node on Earth.

Only 3 companies can do this. TSMC spent $165 billion on their Arizona cluster alone. It took decades of institutional knowledge.

Tesla has manufactured zero semiconductors. Ever.

But they don’t really have a choice choice:
Even in the best-case scenario where TSMC, Samsung, and Micron all max out production for Tesla, it’s still not enough chips for the Optimus and Cybercab roadmap…let alone the rest of the industry

The AI5 chip this fab is designed to produce has 50x the compute of AI4. They need 100-200 billion of them per year. No foundry on the planet can commit to that volume on Tesla’s timeline.

If it works, Tesla becomes one of maybe 4 entities on Earth that can produce frontier AI silicon in-house, and the only one that also builds the robots and cars. It would be game over

[Stitch by Google @stitchbygoogle]

Meet the new Stitch, your vibe design partner.

Here are 5 major upgrades to help you create, iterate and collaborate:
🎨 AI-Native Canvas
🧠 Smarter Design Agent
🎙️ Voice
⚡️ Instant Prototypes
📐 Design Systems and DESIGN.md

Rolling out now. Details and product walkthrough video in 🧵

[Josh Kale @JoshKale]

The AI industry is about to run out of chips. The world's largest foundries are maxed out and the demand curve is still going vertical.

Tesla’s TeraFab is a bet on a solution no other company in the world has been ambitious enough to try.... build the chips yourself

Quick math on how big this is:
- Targeting 70% of TSMC’s global output
- 2nm chips. In-house design
- 1/10th the cost of Nvidia Blackwell
- 1/3rd the power consumption
- $20B capex budget for 2026
- Launchin next week

Elon said on the earnings call that even with all fabs fully booked out, the best-case chip supply projections aren't enough. He said there are zero advanced memory fabs at scale in the US. Literally zero.

So he's building the entire stack. Chips. Memory. Packaging. Energy. All domestic.