White House issues executive order on AI security risks

Microsoft built an AI agent called Scout that lives in Teams and handles your calendar, email, and messages while you're not working.

It's built on OpenClaw — the same tech that blew up in San Francisco earlier this year. Scout shows up in Teams like a regular coworker. You message it tasks, it does them.

What it can do: reschedule meeting conflicts, draft replies, track promises people make to you (and you make to others), send follow-up reminders, and block off personal time on your calendar.

It's rolling out to a limited group first. There's also a desktop app for users with Frontier access, though it requires a GitHub Copilot subscription.

It's not perfect. The VP in charge of the product said his own Scout sent an email as one long run-on sentence with no formatting.

Security risk worth knowing: agentic tools like this are vulnerable to prompt injection — where bad actors manipulate the bot into doing or revealing things it shouldn't. Microsoft says it's tracking everything agents do.

Google has a competing product called Gemini Spark, announced at Google I/O. Same idea, different brand.

The pattern is clear: after coders, office workers are next in line for the agentic treatment. Logistics, scheduling, internal comms — Scout handles it around the clock, whether you're in the office or not.



𝗣𝗿𝗲𝘀𝗶𝗱𝗲𝗻𝘁 𝗧𝗿𝘂𝗺𝗽 𝗔𝗱𝗷𝘂𝘀𝘁𝘀 𝗦𝗲𝗰𝘁𝗶𝗼𝗻 𝟮𝟯𝟮 𝗠𝗲𝘁𝗮𝗹𝘀 𝗧𝗮𝗿𝗶𝗳𝗳𝘀 𝘁𝗼 𝗦𝘂𝗽𝗽𝗼𝗿𝘁 𝗨.𝗦. 𝗠𝗮𝗻𝘂𝗳𝗮𝗰𝘁𝘂𝗿𝗶𝗻𝗴 𝗮𝗻𝗱 𝗜𝗻𝘃𝗲𝘀𝘁𝗺𝗲𝗻𝘁

President Trump has signed a new Proclamation modifying certain Section 232 steel and aluminum tariffs with the stated goal of strengthening U.S. national security, encouraging domestic investment, and supporting key industries such as agriculture, housing, and manufacturing.

Key changes include:

🔹 Tariffs on certain agricultural equipment, including combines and harvesters, as well as other designated equipment, 𝘄𝗶𝗹𝗹 𝗯𝗲 𝗿𝗲𝗱𝘂𝗰𝗲𝗱 𝗳𝗿𝗼𝗺 𝟮𝟱% 𝘁𝗼 𝟭𝟱%

🔹 The existing category of industrial equipment eligible for a 15% tariff rate has been expanded to include certain mobile industrial equipment, such as bulldozers, forklifts, and similar machinery, when imported from qualifying trade agreement countries.

🔹 Foreign manufacturers are incentivized to increase U.S. metal sourcing through a new provision allowing certain capital equipment to qualify for a 10% duty rate if it contains at least 85% U.S. melted and poured steel or U.S. smelted and cast aluminum by weight.

🔹 These measures are temporary and are scheduled to remain in effect through December 31, 2027, with the objective of driving near-term investment and helping rebuild the U.S. industrial base.

TRUMP AI EO

Today, President Trump signed a significant Executive Order on artificial intelligence titled "Promoting Advanced Artificial Intelligence Innovation and Security." This order aims to assert national security oversight over the most powerful AI systems while avoiding the regulatory framework that Trump declined to sign just two weeks ago.

The centerpiece of the EO is a voluntary pre-release review framework for "covered frontier models," which are the most advanced AI systems designated by the NSA through a classified benchmarking process. Developers may voluntarily provide the government with up to 30 days of access to their most powerful models before broader release. This comes in the wake of Anthropic’s Mythos Preview model, which showcased an impressive ability to autonomously identify and exploit software vulnerabilities in critical systems, leading to emergency meetings among Treasury Secretary Bessent, outgoing Fed Chair Powell, and Wall Street CEOs earlier this spring.

The EO also mandates sweeping cybersecurity hardening measures on short deadlines, including the establishment of a new AI cybersecurity clearinghouse co-managed by Treasury, NSA, and CISA to coordinate vulnerability scanning and patch distribution.

Two features of the EO deserve particular legal scrutiny. First, while participation in the voluntary program is optional, the designation of a "covered frontier model" is made unilaterally by the NSA through a classified process that companies cannot observe, challenge, or appeal. As one critic noted: "You can decline the program. You can’t decline the label." Second, the EO includes an explicit anti-licensing carve-out, ensuring that nothing authorizes mandatory preclearance or permitting for AI model releases, which was crucial for securing the President’s signature.

Notably absent from the EO are civil rights or algorithmic fairness provisions, mandatory reporting obligations, federal preemption of state AI law, and any private right of action. For employers and AI governance practitioners, the existing state regulatory patchwork — including laws from Illinois, California, Colorado, and New York — remains fully operative and continues to be the primary compliance frontier.

🔥 Microsoft is going all-in on the AI future!

At their Build 2026 developer conference, Microsoft just dropped some major announcements that could reshape how we use computers:

 🚀 Highlights:

- **Project Solara**: A new line of AI-powered devices (think smart speaker or keycard size) that **don’t run traditional apps or operating systems**. Instead, they host intelligent AI agents that handle real tasks — like documenting medical visits — by connecting to the cloud.

- **Surface RTX Spark Dev Box**: A beast of a PC powered by Nvidia’s new chip. CEO Satya Nadella called it a "**dream machine**" and said he’s on the waitlist himself.

- **Safer OpenClaw on Windows**: Microsoft is making the popular open-source AI agent tool safe for enterprise use, so companies can run powerful AI agents without risking sensitive data.

- **New AI Agent “Scout”**: Inside Copilot, this agent will help manage emails, messages, and decisions for you.

- **New AI Models**: Microsoft’s MAI unit released **MAI Thinking-1**, a strong reasoning model that matches top competitors, plus a highly efficient transcription model and image generation tech.

They’re also partnering with the **Mayo Clinic** to build advanced AI for faster, better medical diagnoses.

Microsoft is clearly betting big on **AI agents + specialized hardware** to move beyond traditional apps and operating systems.

Prime Day is coming early this year. Amazon will kick off its savings bonanza on June 23, offering new discounts every day through June 26. The e-commerce giant moved up the event to avoid overlap with major July occasions like the height of FIFA World Cup and the 250th anniversary of U.S. independence. Prime Day 2026 comes as Americans feel increasingly squeezed by high prices. In a nod to shoppers' stressors, Amazon will be focusing on discounted groceries and household necessities.

Anthropic announced Tuesday that 150 more organizations across over 15 countries will test its new Mythos AI model for security vulnerabilities, expanding an initiative dubbed Project Glasswing. The new partners come from industries that were previously under-represented, including power and healthcare, Anthropic said. The company also noted that over 10,000 high or critical security flaws have been exposed since an April rollout of initial testing that included 50 partners. The news comes a day after the Claude maker filed confidentially for an initial public offering.

Victoria's Secret on Tuesday reported strong fiscal first quarter results, with a net income of $47.7 million and a 15% annual increase in sales to $1.56 billion. The company raised its full-year sales guidance, citing lower tariffs and increased customer spending, particularly among younger shoppers. CEO Hillary Super highlighted the success of the Pink line and fewer promotions as key factors. The retailer said it aims to strengthen its brand identity and expand its beauty offerings moving forward.

🚨 Young & unemployed? It's not AI — it's REMOTE WORK. 🚨

We keep hearing that AI is stealing jobs from college grads… but a new study from the Federal Reserve Bank of New York says that's NOT the real culprit.

The real problem? Remote work. 💻🏠

Here's what they found 👇

📈 Unemployment for young college grads in "remotable" jobs (like software dev) jumped ~1 percentage point from 2017-2019 to 2022-2024.

📉 But for workers aged 29+ in those SAME fields? Unemployment actually dropped.

⚠️ In non-remote jobs (like nursing)? No gap between young and old workers.

So what's happening?

👉 Companies don't want to hire fresh grads for remote roles because it's HARDER to train and mentor them from a distance. They'd rather hire experienced workers who need less hand-holding.

The study says remote work is responsible for nearly TWO-THIRDS of the rise in youth unemployment since the pandemic.

And this started before ChatGPT even existed. 🤖

When a Fortune 500 tech company reopened its offices? They started hiring younger workers again. But for teams still working remotely? They kept favoring experienced hires.

So next time someone blames AI for your job search struggles… tell them it's the Zoom call. 😅

🚨 **META IS BACKING DOWN on its controversial employee tracking software!** 🚨


After massive pushback from staff who dubbed the company an “Employee Data Extraction Factory,” Meta is officially dialing back its plan to track employee mouse movements, clicks, and keystrokes to train its AI models. 


**Here’s what’s changing after the employee revolt:**

✅ **Pause Button:** Employees can now pause data collection for up to 30 minutes at a time.

✅ **Opt-Outs:** Workers can now request full exemptions from the tracking.

✅ **Tech Fixes:** Meta is rolling out battery and data optimizations after the software drained laptops and caused employees' home Wi-Fi usage to spike.

Meta originally installed the software to gather data to build autonomous AI agents. But employees quickly revolted over privacy concerns, personal data on work devices, and massive battery drain. This internal rebellion could also deepen Meta’s ongoing regulatory headaches in the EU over data privacy.


Post a Comment

Previous Post Next Post