Hey

Welcome to The AI Report — the #1 AI newsletter for business leaders.

Before today's stories — this is the one to act on: the AI Executive's Pass is live right now. We've been teasing this for a while, and it's finally here. One $199/year membership that bundles $11,000+ in AI tools and resources — Notion, beehiiv, ListKit, Make.com, Paperform, Intercom, a private GPT training session, and a LinkedIn growth consult. Launch pricing only runs until May 19, so if you've been on the fence, now's the time.

Get the AI Executive Pass →

You don’t need to be technical. Just informed.

• 1. 📧 Craft the perfect marketing email with Brew.new 
• 2. ⚡️ Hackers crack Claude Mythos
• 3. 💼 Your Business Briefing
• 4. 📋 Strengthen your compliance with Proton Mail
• 5. ✍️ Today’s Policy Corner
• 6. 🗞️ The News Bulletin

Are you ready to launch AI transformation for your company?

TOGETHER WITH BREW.NEW

Turn Every Signup into Revenue

You're signing up new users but not welcoming, educating, or nurturing them. Your emails are outdated, off-brand, and every new campaign takes days to execute. You're leaving money on the table.

Claude Design, Google Stitch, and ChatGPT can now generate marketing emails. The problem: they break in real inboxes. Gmail, Outlook, and Apple Mail all render HTML differently.

Brew.new solves this.

Describe your campaign and its AI builds the entire sequence in seconds, welcome flows, drip campaigns, newsletters, multiple on-brand variants. Ready to send.

• Renders perfectly across every inbox. Gmail, Outlook, Apple Mail - every email lands exactly as designed

• Pulls your brand automatically from your website and Figma, capturing fonts, colors, imagery, and voice down to the small details

• Agent-native. Claude, Lovable, Replit, and OpenClaw agents can operate Brew straight out of the box

Brew is like Claude Design specifically for Email Marketing!

Latest in AI

Hackers guess Anthropic's new URL

🚨 Our Report

Anthropic's Claude Mythos Preview, its most capable AI model for cybersecurity tasks, was accessed without authorization on the same day the company announced it. A private group combined credentials from a third-party contractor with information from a data breach at AI-recruitment startup Mercor, then guessed the Mythos endpoint URL format to gain entry.

🔓 Key Points

• The unauthorized users exploited a contractor credential and a URL pattern guess to reach the model through a third-party vendor environment, not through any direct Anthropic system breach.

• Anthropic confirmed it is investigating but said it found no evidence of activity beyond the vendor environment. The company will not release Mythos publicly; prediction markets assign just 7% odds to a launch by June 30.

• On ExploitBench, Mythos achieved full arbitrary code execution on 21 of 41 tested vulnerabilities. GPT-5.5 reached that tier on just two. Researchers say Mythos works like "a fairly competent browser / JS engine security researcher."

🔐 Relevance

The breach exposes a core tension for enterprises deploying frontier AI: access controls failed at the supply chain's weakest point, a contractor credential, not through sophisticated hacking. Anthropic's own security team found no evidence of activity beyond the vendor environment, but the fact that access was possible on day one raises questions about how other AI vendors manage third-party permissions. For organizations evaluating AI partnerships, this is a reminder that your exposure extends beyond your own systems. Auditing contractor access and credential hygiene should move up the priority list.

The AI Report Podcast

The SaaSpocalypse: What AI Kills and What Survives | Jason Li, Laurel

The AI Executive's Pass is live

Most business leaders are already paying for 4–6 AI tools they found separately, at full price, with no coordination between them. The AI Executive's Pass is the fix.

One $199/year membership. $11,000+ in tools and resources built for executives actually implementing AI — not just reading about it.

Inside: Notion, beehiiv, ListKit, Make.com, Paperform, Intercom, a private GPT training session, and a LinkedIn growth consult.

Launch pricing ends May 19. Get it while it's here.

THE BUSINESS BRIEFING: SALES (powered by Upscaile)

Allegis Group (global workforce solutions, 18K employees) was drowning in manual compliance work. Their 17-person compliance team managed 85+ weekly processes, analysts spent 30 minutes calculating each PTO accrual, and translation of 1,400 documents cost $900K annually with 3-4 day turnarounds.

They deployed Microsoft 365 Copilot with Azure AI Services to build a PTO Calculator agent (instant accrual calculations) and ALTA translation assistant (minutes vs. days). Copilot Studio handled natural language queries across unstructured compliance data.

Tool used: Microsoft 365 Copilot + Azure AI Services – enterprise AI platform for workflow automation and data analysis.

Result: 150K hours saved across 18K users. PTO case closure dropped from 31 hours to 13 with 100% accuracy across 877 accruals. Translation delivered $1.5M savings, processing 2,000+ documents at 90-95% accuracy in minutes instead of days.

The lesson: Fear blocks adoption faster than technology. Allegis Group invested in foundational training first – one skeptical compliance manager became their biggest champion after seeing the PTO Calculator in action.

Steal this: Identify one repetitive calculation your compliance or ops team does manually. Upload three examples to ChatGPT or Claude with the prompt: "Build me a step-by-step process to automate this calculation." Test it on 10 real cases this week before rolling it wider.

TOGETHER WITH PROTON

How do you show customers that you take their security seriously? 

Whether you're in healthcare, finance, or any other highly regulated industry, your clients care about privacy. Proton Mail strengthens your GDPR, HIPAA, PCI, and SOC 2 compliance because our end-to-end encryption and zero-access architecture meet strict regulatory standards by default. 

You never have to "opt in" to security. It's built into every email you send.

THE POLICY CORNER

California hits GM with record $12.75M fine for selling driver data without proper consent

General Motors and OnStar settled with California regulators over allegations they collected and sold connected vehicle data—precise geolocation, braking patterns, speed, trip timing, and seatbelt usage—from hundreds of thousands of California drivers to data brokers LexisNexis and Verisk without adequate notice or opt-out mechanisms. GM earned roughly $20M from these sales nationwide. The case marks the largest California Consumer Privacy Act penalty to date and represents the first enforcement of CCPA's data minimization and purpose limitation rules.

Deadline: In effect now. Enforcement began May 8, 2026.

Your move: Audit any data flows to third-party analytics providers or brokers—confirm your opt-out mechanisms actually stop all applicable data sales, not just some. If you collect precise geolocation or behavioral data, document a completed risk assessment and verify your privacy team has reviewed it at the executive level.

AI News

• 🔬 Researchers build magnetic switch 1,000x faster than modern AI chips: University of Tokyo team demonstrates picosecond-speed switching using antiferromagnet material, generating minimal heat and using fraction of energy compared to current accelerators. FULL STORY

• 📱 Google I/O 2026 preview points to Gemini 4.0 and Android XR glasses launch: Pre-show leaks reveal seven new Gemini Live voice models under testing, plus confirmation that display-equipped Android XR smart glasses will debut at May 19 event. FULL STORY

• 😰 Meta employees brace for 8,000 layoffs amid AI-driven restructuring: Staff report anxiety and exhaustion as notifications expected May 20, with morale impacted by falling compensation, mandatory activity tracking software, and pressure to build AI systems that may replace their roles. FULL STORY

• ⛪ Vatican forms AI ethics commission to guide Catholic Church's technology stance: Pope announces specialized team including tech experts, philosophers, and ethicists to develop frameworks addressing privacy, warfare applications, job displacement, and ensuring AI serves human dignity. FULL STORY

Trending AI Tools (Sponsored by the AI Leaders Guide)

A curated look at the AI tools quietly transforming how teams work.

1. Thenvoi* - The infrastructure that lets your AI agents collaborate across any framework.

2. Animatives is an app for crafting stop motion and timelapse animations with frame-by-frame precision

3. Flockx is a collaboration framework that integrates AI agents with AI marketplaces

^{*indicates a sponsored tool, if any}

⚡️ Looking for the exact AI tools our team uses to run The AI Report?

The Money: Holding companies bet on data infrastructure

Publicis is acquiring LiveRamp for $2.5B at a 30% premium, positioning data collaboration as the substrate for enterprise AI agents. The deal follows a pattern: agencies are buying the pipes that feed autonomous systems, not just the models that run them.

Deals to know:

• Publicis Groupe (Acquisition, $2.5B) -- Acquires LiveRamp's clean rooms, identity infrastructure, and 21,000+ publisher domains to enable "data co-creation" across client partners. Adds to 2019 Epsilon buy and March 2025 Lotame acquisition. Investors: All-cash transaction funded by Publicis balance sheet and debt

• Cerberus (IPO, $5.6B) -- AI chipmaker debuts as largest 2026 public offering. Shares surged 68% on launch day before dropping 10% Friday. Part of a $28.4B year-to-date US IPO wave centered on AI infrastructure, aerospace, and energy. Investors: Public markets

Signal: Capital is flowing into infrastructure that makes AI agents functional, not just smarter. Holding companies and chipmakers are positioning themselves as essential middleware between models and outcomes.

Refer a Friend

Latest episode: The SaaSpocalypse: What AI Kills and What Survives → Listen here
Want to reach 400k+ decision-makers? → Sponsor us

Until next time, Arturo and Liam.

P.S. Unsubscribe if you don’t want us in your inbox anymore.